Det här verket har digitaliserats vid Göteborgs universitetsbibliotek. Alla tryckta texter är OCR-tolkade till maskinläsbar text. Det betyder att du kan söka och kopiera texten från dokumentet. Vissa äldre dokument med dåligt tryck kan vara svåra att OCR-tolka korrekt vilket medför att den OCR-tolkade texten kan innehålla fel och därför bör man visuellt jämföra med verkets bilder för att avgöra vad som är riktigt. Th is work has been digitised at Gothenburg University Library. All printed texts have been OCR-processed and converted to machine readable text. Th is means that you can search and copy text from the document. Some early printed books are hard to OCR-process correctly and the text may contain errors, so one should always visually compare it with the images to determine what is correct. 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 C M !THE COMBINED APPLICATION OF BIBLIOGRAPHIC COUPLING AND THE COMPLETE LINK CLUSTER METHOD IN BIBLIOMETRIC SCIENCE MAPPING BO JARNEVING VALFRID THE COMBINED APPLICATION OF BIBLIOGRAPHIC COUPLING AND THE COMPLETE LINK CLUSTER METHOD IN BIBLIOMETRIC SCIENCE MAPPING BO JARNEVING Akademisk avhandling som med tillstånd av samhällsvetenskapliga fakulteten vid Göteborgs universitet för vinnande av doktorsexamen framläggs till offentlig granskning kl 13.15 fredagen den 10 februari 2006 i Stora hörsalen (C203), Högskolan i Borås, Allégatan 1, Borås. Institutionen Biblioteks- och informationsvetenskap/Bibliotekshögskolan Högskolan i Borås och Göteborgs universitet Title: The combined application of bibliographic coupling and the complete link cluster method in bibliometric science mapping Abstract: This thesis connects to previous research in bibliometric science mapping and citation indexing. A method was suggested for science mapping purposes and evaluated. The suggestion of this method was motivated by the fact that the prevailing method of citation based science mapping of documents, the cocitation cluster analytical method, can not map the most current published research, a feature that is a characteristic of the proposed method. On theoretical grounds, it was assumed that neither of these methods could substitute for the other and that they would have complementary functions in relation to one another. The prime objective was to evaluate the proposed method’s capability to generate subject coherent clusters, i.e. to identify coherent research themes, and the assumed context of application was scientific information provision. The proposed method has two primary components: ( 1 ) a measure of document similarity, bibliographic coupling and (2) a cluster analytical method for the partition of document populations, the complete link cluster method. The research design comprised four different research settings of which three correspond to specific fields of research and one to a large multidisciplinary environment. Methods of evaluation comprised quantitative approaches as well as more qualitative ones. For the establishment of cluster coherence, measures of density and average coupling strength in clusters were applied. The relevance of generated clusters was assumed to be reflected by these measures and was substantiated by field experts’ evaluations ofclustering results. In order to assess the agreement between field experts’ apprehensions of their fields’ cognitive structures, intellectual-manual partitions of document populations were performed by field experts and compared with partitions generated by the proposed method. Findings showed that the proposed method has the capability to identify and map current and coherent research themes on the level of a single research field as well as in a multidisciplinary environment. However, based on theoretical considerations as well as on empirical findings, it was concluded that it would not suffice as a standard science mapping method where exhaustive depictions of specialties’ cognitive structures are aimed at. The reasons for this were: i. As for now, the method of bibliographic coupling can not identify the most central concepts of a research specialty. ii. The dependency of consensual referencing implies that only minor shares of original document populations will be available for analysis. iii. The lack of a method for the decision of appropriate thresholds of coupling strength implies arbitrary threshold settings. iv. The partition of document populations brought about a fragmentation of research fields. v. Partitions generated by field experts deviated considerably from partitions generated by the complete link cluster method. It was therefore concluded that the proposed method may be complementary to the cocitation cluster analytical method and to traditional citation indexing. Based on the empirical findings, a tentative outline for such an application was given. Keywords: bibliometrics, bibliographic coupling, science mapping, citation indexing, cocitation analysis, cluster analysis, scientific information provision THE COMBINED APPLICATION OF BIBLIOGRAPHIC COUPLING AND THE COMPLETE LINK CLUSTER METHOD IN BIBLIOMETRIC SCIENCE MAPPING BO JARNEVING Akademisk avhandling som med tillstånd av samhällsvetenskapliga fakulteten vid Göteborgs universitet för vinnande av doktorsexamen framläggs till offentlig granskning kl 13.15 fredagen den 10 februari 2006 i Stora hörsalen (C203), Högskolan i Borås, Allégatan I, Borås. Institutionen Biblioteks- och informationsvetenskap/Bibliotekshögskolan Högskolan i Borås och Göteborgs universitet Title: The combined application of bibliographic coupling and the complete link cluster method in bibliometric science mapping Abstract: This thesis connects to previous research in bibliometric science mapping and citation indexing. A method was suggested for science mapping purposes and evaluated. The suggestion of this method was motivated by the fact that the prevailing method of citation based science mapping of documents, the cocitation cluster analytical method, can not map the most current published research, a feature that is a characteristic of the proposed method. On theoretical grounds, it was assumed that neither of these methods could substitute for the other and that they would have complementary functions in relation to one another. The prime objective was to evaluate the proposed method’s capability to generate subject coherent clusters, i.e. to identify coherent research themes, and the assumed context of application was scientific information provision. The proposed method has two primary components: ( 1 ) a measure of document similarity, bibliographic coupling and (2) a cluster analytical method for the partition of document populations, the complete link cluster method. The research design comprised four different research settings of which three correspond to specific fields of research and one to a large multidisciplinary environment. Methods of evaluation comprised quantitative approaches as well as more qualitative ones. For the establishment of cluster coherence, measures of density and average coupling strength in clusters were applied. The relevance of generated clusters was assumed to be reflected by these measures and was substantiated by field experts' evaluations of clustering results. In order to assess the agreement between field experts' apprehensions of their fields’ cognitive structures, intellectual-manual partitions of document populations were performed by field experts and compared with partitions generated by the proposed method. Findings showed that the proposed method has the capability to identify and map current and coherent research themes on the level of a single research field as well as in a multidisciplinary environment. However, based on theoretical considerations as well as on empirical findings, it was concluded that it would not suffice as a standard science mapping method where exhaustive depictions of specialties’ cognitive structures are aimed at. The reasons for this were: i. As for now. the method of bibliographic coupling can not identify the most central concepts of a research specialty. ii. The dependency of consensual referencing implies that only minor shares of original document populations will be available for analysis. iii. The lack of a method for the decision of appropriate thresholds of coupling strength implies arbitrary threshold settings. iv. The partition of document populations brought about a fragmentation of research fields. v. Partitions generated by field experts deviated considerably from partitions generated by the complete link cluster method. It was therefore concluded that the proposed method may be complementary to the cocitation cluster analytical method and to traditional citation indexing. Based on the empirical findings, a tentative outline for such an application was given. Keywords: bibliometrics, bibliographic coupling, science mapping, citation indexing, cocitation analysis, cluster analysis, scientific information provision THE COMBINED APPLICATION OF BIBLIOGRAPHIC COUPLING AND THE COMPLETE LINK CLUSTER METHOD IN BIBLIOMETRIC SCIENCE MAPPING BO JARNEVING VALFRID 2005 THE COMBINED APPLICATION OF BIBLIOGRAPHIC COUPLING AND THE COMPLETE LINK CLUSTER METHOD IN BIBLIOMETRIC SCIENCE MAPPING BO JARNEVING DOCTORAL THESIS DEPARTMENT OF LIBRARY AND INFORMATION SCIENCE/SWEDISH SCHOOL OF LIBRARY AND INFORMATION SCIENCE UNIVERSITY COLLEGE OF BORÅS/GÖTEBORG UNIVERSITY THE COMBINED APPLICATION OF BIBLIOGRAPHIC COUPLING AND THE COMPLETE LINK CLUSTER METHOD IN BIBLIOMETRIC SCIENCE MAPPING BO JARNEV1NG Distribution: The Publishing Association Valfrid Department of Library and Information Science/Swedish School of Library and Information Science University College of Borås/Göteborg University Copyright: The Author and Valfrid Print: Intellecta Docusys, 2005 Series: Publications from Valfrid, nr 30 ISBN 91-89416-12-0. ISSN 1103-6990 ACKNOWLEDGEMENTS I wish to thank Olle Persson and Elena Maceviciuté for their supervision of this thesis. Richard Danell Anders Kastberg Göran Levan Peder Svensson Länghem, 2005 I would also like to thank the following researchers for taking time off their busy schedules to evaluate the relevance of the mapping results of the empirical studies: 1 also wish to express my gratitude to Per Ahlgren for the many fruitful discussions and good advice and to Ronald Rousseau for his good advice and suggestions. Many thanks also to Johan Eklund for providing me with the much needed technical support and for programming. Further, thanks to Boel Bissmarck for checking the English and to Christian Swalander for editing. Bengt Alrud Kim Bolton Bo Jameving TABLE OF CONTENTS Chapter 1 Introduction 11 Chapter 2 The Theoretical Framework 15 1. Central Concepts 15 1.1 Citation indexing 15 1.2 Citation analysis 16 1.2.1 Basic assumptions underlying citation analysis 17 1.2.2 Problems of citation data and sources 19 1.2.3 Citation based science mapping 21 1.3 Mathematical concepts and definitions 23 1.4 Classification and cluster analysis 25 1.4.1 Cluster analytical methods 26 1.4.2 Motives for the choice of the complete link cluster method 29 2. Previous Research 31 2.1 Bibliographic coupling 31 2.2 Cocitation analysis 36 3. Summary and Foundation for the Research Design 43 3.1 Origination of methods and direction of development 43 3.2 Comparison of properties of methods 43 3.3 Presumed general problems of citation based document mapping 44 3.4 Methods of partition 46 3.5 A foundation for the research design 48 Chapter 3 Rationale and Research Design 50 1. Research Settings 50 2. Rationale and Research Questions 50 2.1 Cases 1 to 3 51 2.2 Case 4 52 Chapter 4 Methods and Data 56 1. The Basic Components of the Proposed Method 56 561.1 Measurement of proximity 1.2 Application of the complete link clu ster method 58 1.3 Application of the between groups average cluster method 61 1.4 Comparison of cluster methods 61 2. Methods of Evaluation 64 2.1 The Qualitative assessment of cluster compositions 64 2.2 The Quantitative assessment of cluster compositions 65 2.3 Comparison of partitions with regard to Cases 1 to 3 66 2.4 The intellectual manual partitions generated by the field experts 68 2.5 Visualization of partitions 68 3. Data Selection, Threshold Setting and Features of Final Populations 69 3.1 Thresholds and observation period 69 3.2 Research Settings 73 3.2.1 Casel 73 3.2.2 Case 2 74 3.2.3 Case 3 75 3.2.4 Case 4 76 Chapter 5 Findings 78 1. Case 1 : Scientometrics 78 1.1 Clusters generated by the complete link cluster Method 78 1.1.1 Coherence and separation 78 1.2 Clusters generated by the field expert 80 1.2.1 The partition 80 1.2.2 Coherence and separation 81 1.3 Analysis and comparison of partitions 83 1.3.1 The coherence of clusters 83 1.3.2 The separation between clusters 83 1.3.3 The concentration of articles to clusters 83 1.3.4 The qualitative assessment of cluster compositions 83 1.4 The field expert’s evaluation 84 1.5 Summary of findings in Case 1 84 2. Case 2: Organic Chemistry 87 2.1 Clusters generated by the complete link cluster Method 87 2.1.1 Coherence and separation 87 2.2 Core documents - a microanalysis 89 2.3 Clusters generated by the field expert 89 2.3.1 The partition 89 2.3.2 Coherence and separation 91 2.4 Analysis and comparison of partitions 93 2.4.1 The coherence of clusters 93 2.4.2 The separation between clusters 93 2.4.3 The concentration of articles to clusters 93 2.4.4 The qualitative assessment of cluster compositions 94 2.5 The field expert’s evaluation 94 2.6 Summary of findings in Case 2 96 3. Case 3 : Pure & Applied Mathematics 99 3.1 Clusters generated by the complete link cluster method 99 3.1.1 Coherence and separation 99 3.2 Clusters generated by the field expert 101 3.2.1 The partition 101 3.2.2 Coherence and separation 103 3.3 Analysis and comparison of partitions 105 3.3.1 The coherence of clusters 105 3.3.2 The separation between clusters 105 3.3.3 The concentration of articles to clusters 105 3.3.4 The qualitative assessment of cluster compositions 105 3.4 The field expert’s evaluation 106 3.5 Summary of findings in Case 3 106 4. Case 4 : Core Documents 109 4.1 The first fusion level - Cl clusters 110 4.1.1 Clusters and cluster sizes 110 4.1.2 Coherence and separation 111 4.1.3 Example of cluster fusion on the C1 level 113 4.2 The second fusion level - C2 clusters 116 4.2.1 Clusters and cluster sizes 116 4.2.2 Coherence and separation 117 4.2.3 Example of cluster fusion on the C2 level 119 4.3 The third fusion level - C3 clusters 122 4.3.1 Clusters and cluster sizes 122 4.3.2 Coherence and separation 123 4.3.3 Example of cluster fusion on the C3 level 125 4.4 Field Experts’ evaluations of 4 cases of iterated Clustering 134 4.4.1 Cluster C3/12: “Human genetics and disease” 134 4.4.2 Cluster C3/19: “Chemistry” 136 4.4.3 Cluster C3/27: “Bose-Einstein Condensation” 139 4.4.4 Cluster C3/29: “Carbon-Nano Tubes” 141 4.5 The expansion of Cl-clusters 144 4.6 Summary 149 Chapter 6 Discussion and Conclusions 1. Discussion 151 151 1.1 Cases 1 to 3 1.1.1 The relevance of clusters ge nerated by the 151 complete link cluster method 1.1.2 The extent and nature of deviations between results generated by the complete link cluster method and results generated by 151 Intellectual manual partitions 1.1.3 A commentary on and comparison of 151 methods of partition 1.1.4 The effects of threshold settings and method of partition on the original populations of 153 research articles 154 1.1.5 Implications of findings 156 1.2 Case 4 1.2.1 The extent of fragmentation imposed by the 157 applied method 1.2.2 The impact of iterated clustering on the 157 overall cluster structure 157 1.2.3 The optimal level of cluster fusion 157 1.2.4 Implications of findings 1.3 Reflections on findings in relation to previous 158 research 158 2. Conclusions 160 References 163 Appendix 1 Equations 167 Appendix 2 Bibliographic descriptions of clusters with a size > 3 in Case 1 171 Appendix 3 The comparison of two partitions in Case 1 176 Appendix 4 Bibliographic descriptions of clusters with a size > 3 in Case 2 178 Appendix 5 The comparison of two partitions in Case 2 197 Appendix 6 Bibliographic descriptions of core document clusters in Case 2 201 Appendix 7 Bibliographic descriptions of clusters with a size > 3 in Case 3 204 Appendix 8 The comparison of two partitions in Case 3 214 CHAPTER 1: INTRODUCTION Bibliometrics is the quantitative study of patterns derived from the production and use of publications. It was defined by Pritchard in 1969 as "the application of mathematical and statistical methods to books and other media of communication". It is most often used in the field of library and information science, but has also wide applications in other areas (e.g. science policy). An important area of bibliometric research is citation analysis. This sub-field comprises several methods for the analysis of citation relations in research literatures. The analysis of citations originates from the need of scientists to build on previous research when embarking on new research projects and to refer back to them when publishing the results. When referring back to previous research, the publishing scientist sets the framework of his research, while the publishing of the research itself can be seen as the individual scientist’s claim of intellectual property and the seeking for acknowledgement by peers. This acknowledgement is in turn reflected by possible future citations in other scientists’ subsequent publications. In Ziman (1984, p. 58) it is stated that: ...the basic principle of academic science is that results of research must be made public /.../. Whatever scientists think or say individually, their discoveries cannot be regarded as belonging to scientific knowledge until they have been reported to the world and put on permanent record. Based on the needs of scientists to find and reference previous published research, so called citation indexes have been constructed. A citation index facilitates the retrieval of documents associated through citation links, and is complementary to other information retrieval methods. 1 The development of citation indexing and the launching of citation databases by the Institute for Scientific Information (ISI) during the ’60s have been fundamental for the development of citation analytical methods, in particular citation based science mapping (Garfield, 1998). 1 “Information retrieval deals with the representation, storage, organization of and access to information items” (Baeza-Yates & Ribeiro-Neto, 1999, p. 1). 2 The Atlas of Science was presented in 198 land was based on the clustering of highly cited and cocited documents from a given sub-specialty and provided the user with a mini-review of the subject, a bibliography of clustered documents, a cluster-map depicting the documents in a cluster - the similarity or distance between them - and a bibliography of documents citing the clustered documents. Citation based science mapping is an area of bibliometrics where the structure and development of science are elaborated and visualized through the analysis of bibliographic data, representing research documents, mostly articles. The objective for citation based science mapping has commonly been to reveal the cognitive structure of science in terms of visualizing and describing its sub-division in disciplines (fields), sub-disciplines and specialties. Also, mappings have been focusing on scientific information provision (e.g. the ISI product Atlas of Science2). The notions of discipline, sub-discipline and specialty should be clarified. A discipline should be the broadest entity, denoting a branch of scholarly knowledge, e.g. physics. Physics in turn can be divided in sub-disciplines like condensed matter physics which in turn can 11 be divided in specialties like solid state physics, materials physics and polymer physics, which in turn can be divided in other (sub-) specialties. These terms reflect a function of continuous specialization, subdivision and new amalgamations of research over time, rather than well demarcated and static hierarchical levels of classification. This function of specialization is due to the fact that a single researcher can not attain a detailed knowledge of all areas within a certain discipline. Hence, by necessity researchers must focus on a specific area within their fields or sub-fields. Those researchers with a common focus communicate (both formally [through academic journals] and informally) and over time such a group with a specialized research focus form an area of specialization. The term “field” is frequently used in the literature and may cover any demarcated area of research.3 3 The general term “field” mostly denotes the discipline level or the specialty level, depending on the context. It is often difficult to classify the exact level of scientific activity and the use of terms in the literature is ambigious and inconsistent. The terms“sub-field” and “sub-specialties” are sometimes used as well. 4 It should be noted that the verb “map” indicates that something is mapped, while the noun “map” stands for a graphical representation that may enhance our spatial understanding of associations between objects. Hence, in the context of science mapping, mapping need not lead to maps, though it often does. In citation based science mapping, different entities (journals, authors or documents) in bibliographic descriptions representing research documents are applied as analyzed units for different purposes. For instance, when mapping4 citation relationships between journals, an overall view of the discipline structure of sc ience may be arrived at. However, the journal is a too broad a unit of analysis to reveal the fine structure of science (Small, 1974). Hence, citation based mapping with the objective to map specialties usually employs documents as the unit of analysis and it has been suggested that the “[s]pecialty is the principal mode of social and cognitive organization in modem science” (Small, 1977). The usefulness of science mapping is clear as “most scientists have intuitive notions about the subdivisions of their fields, but no observer, however broadly trained, can gain an overall perspective in the scientific mosaic” (Small, 1974). The difficulty for researchers to gain an intellectual key map over their own discipline’s subdivision in specialties and research foci within specialties is augmented by the increasingly interdisciplinary character of research where new lines of research transcend boarders between disciplines. A good example of this is the (non-traditional) “field” of environmental science, which connects several disciplines and sub-disciplines like astrophysics, chemistry, ecology etc. Conclusively, the mapping of research specialties may provide means, not only for the study of the specialty structure of science, but also for new approaches of indexing and information provision for scientists (cf. Small, 1973). Historically, the development of citation based science mapping is associated with experiments that were launched in the ’70s by I SI where the mapping method was cocitation cluster analysis. This method is defined by the measure of document similarity and the method of clustering applied. The measure of document similarity is the cocitation of documents and single link clustering is the cluster method. Though several improvements of the cocitation cluster technique have been accomplished over the years, the method of document cocitation clustering has been criticized on 12 methodological grounds (Leydesdorff, 1987; Oberski, 1988). The advocates of this method claim that the fine structure of science in terms of identified and mapped specialties is reflected. This has been seriously questioned on grounds of statistical instability resulting from arbitrary application of threshold settings and the use of the single link cluster method. In spite of the criticism, the basic application of document cocitation clustering has not changed and is still dominating as at today. On the other hand, there exists another citation based measure of document similarity, namely, bibliographic coupling, which was introduced to the research community in the early ’60s (Kessler, 1962 and 1963a). In comparison with the cocitation approach, bibliographic coupling methods have the advantage of being capable of identifying emerging specialties (Glänzel & Czerwon, 1995 and 1996), as research articles are available for analysis as soon as they are published. In the case of cocitation analysis, there will always be a time lag between the current published research and the generation of a sufficient number of received citations that can facilitate stable sets of cocitation data for mapping. However, there is also another distinct difference between the cocitation and bibliographic coupling approaches. With regard to cocitation, claims of the identification and mapping of research specialties is based on the presumption that highly cited documents represent central concepts of specialties and that the grouping of such highly cited items on basis of cocitation therefore would reflect the cognitive structures of specialties. With regard to bibliographic coupling, claims can generally not be made that articles represent central concepts as no immediately applicable criterion for this exists. Hence, applying bibliographic coupling for mapping purposes, one could generally not make the same claim of identifying the cognitive structure of a research specialty. This means that cocitation analysis and bibliographic coupling should be complementary to each other. Despite its favorable features, there is a distinct lack of evaluative research concerning bibliographic coupling applied as a science mapping method. The reasons for this unobtrusive position in science mapping are not obvious and comparable and complementary results to the cocitation approach have also been reported when this measure was applied for science mapping purposes (Sharabchiev, 1988; Persson, 1994; Jarneving, 2001). In addition, research in bibliographic coupling has shown that the identification of “hot” research areas could be accomplished by the identification of “core documents”, i.e. currently published research articles with many and strong associations of bibliographic coupling to other currently published research articles, and that most core documents belong to a few high impact documents of a specialty (Glänzel & Czerwon, 1996). For citation based science mapping in general, it also holds that only a small fraction of articles of a selected original population is available for mapping as citation based science mapping depends on consensual referencing. This means that a lack of consensus about which previous research that is the most significant in relation to a common topic, or less attentive referencing, would lead to a loss of cognitive association between articles and a diminishing of the original population (cf. Braam, Moed & van Raan, 1991). This concerns the extent of exhaustiveness of mapping results and affects the validity of claims of identification and mapping of specialties. Conclusively, citation based science mapping is generally attached with uncertainty when the objective is set to identify and define the specialty structure of science. With 13 regard to information provision or information sharing objectives, this uncertainty should have lesser importance as the currency and relevance of obtained information should be the first priority, not the exactness of the mirroring of specialty cognitive structures. Based on the findings of the various researches so far, bibliographic coupling could be combined with a cluster method to provide a method of science mapping complementary to the prevailing cocitation cluster analytical method. The complete link cluster method would on theoretical grounds (cf. Everitt, Landau & Leese, p. 60- 62) provide a suitable cluster method for this purpose, for more coherent clusters would be generated, meaning that it would not have the drawbac ks of the single link cluster method. Thus, based on empirical evidence and theoretical considerations, bibliographic coupling and the complete link cluster method were: combined to form a method of science mapping which was then evaluated in this study. The objective was set to evaluate the proposed method’s capability to generate subject coherent clusters, i.e. to identify coherent research themes, and the assumed context of application was scientific information provision. The research design comprised four different research settings of which three correspond to specific fields of research and one to a large multidisciplinary research setting, where the specific objective was to identify and apply core documents for the evaluation of the applicability of the proposed method. Conclusively, the method to be evaluated has the following two primary components: i. a measure for the association of documents where the association can be expressed as the similarity between two documents; and ii. a cluster analytical method for the partition of sets (populations) of documents. The measure of document similarity is needed for the purpose of establishing cognitive relationships between documents. The cluster method is needed for the partition of a set of documents into subsets of reciprocally similar documents. In this study, bibliographic coupling is applied as the measure of document similarity and the complete link cluster method is used for the clustering of docu ments. The whole research process and its findings are presented in five subsequent chapters beginning with Chapter 2, in which the framework of the thesis is presented. In Chapter 3, the research design, the rationales and the research questions are given. Chapter 4 presents bibliometric and statistical methods applied in this study, the methods of data selection and collection as well as the properties of the data collected. Chapter 5 sets out the findings of the study whilst Chapter 6 discusses the findings and gives the conclusions. In order to facilitate the reading, a list of equations discussed in the thesis is given in Appendix 1. 14 CHAPTER 2: THE THEORETICAL FRAMEWORK In this chapter, the framework on which the design of the study is based is accounted for. It begins with an elaboration of some concepts which are central to the study. Next, the previous research on which the study builds is presented where the outline of the development of cocitation analysis and bibliographic coupling is given. The purpose of presenting both methods is foremost due to the claim made in this thesis that the proposed method would be complementary to the cocitation cluster analytical method. Another motive is that little empirical experience exists concerning bibliographic coupling in the context of science mapping, whereas the development of cocitation analysis follows a clearly discernable track with a series of connected articles on science mapping. This means that experience of citation based science mapping on the document level must be derived from empirical findings from cocitation analysis. The chapter ends with a summary and a discussion of the foundation for the research design of this study. 1. CENTRAL CONCEPTS 1.1 Citation Indexing Citation indexing was developed as a result of the needs of scientists to find and reference previous published research. A citation index lists documents that have been cited and identifies the sources of the citation. The strength of citation indexing is its simplicity. Just by knowing an item that has been cited, several additional documents can be found. Semantic difficulties are avoided as citation symbols rather than words are used to describe the content of a document. This makes the job of the researcher easier when searching for works from other disciplines, as they are not required to know the terminology of the disciplines that they are searching in order to make the search. Traditional subject indexing involves specialist judgment, increasing the time and the cost of indexing with increasing indexing depth5. Citation indexing solves the depth versus cost problem by substituting the author’s citations for the indexer’s judgments and there are no restrictions as to the number of citations (the reason why citation indexing in most cases should be deeper than subject indexing where a few' indexing terms are used). Also important is that citations are timeliness, whereas the usability of an indexing term is due to semantic stability meaning that the actuality of indexing terms might be low in subject indexes, thus, limiting their effectiveness as search tools (Garfield, 1979, p.l). 5 “Indexing depth” aims at the degree to which a topic is represented in detail. In 1961, the database publishing company ISI started to publish the Science Citation Index (SCI) and in 1966 it publishes the Social Science Citation Index (SSCI). The SCI provides access to 3,700 technical and science journals and the SSCI covers 1,700 social science journals. In 1976, subsequently, ISI 15 started to publish the Arts and Humanities Citation Index (A & HCI), which provides access to 1,130 arts & humanities journals. It should be noted that the ISI databases are multidisciplinary, whereas traditional indexing and abstracting services provide databases that are limited to a single field. The SCI and the SSCI have consistently been used by the vast majority of research that applies citation based mapping techniques. The A & HCI has also been used but to a considerably lesser extent. Citation data is made accessible either by downloading hundreds or thousands of bibliographic records from citation databases, or through online techniques (cf. Persson, 1988). In this study, data from the SCI and the SSCI are used. 1.2 Citation Analysis Citation analysis is the area of bibliometrics which deals with the study of the relationships between items of the scientific literature. Several areas of the successful applications of citation analysis have been developed. They include science mapping, information retrieval (IR), evaluation of scientific activity, collection management and history of science. Below is a brief description of these areas of application of citation analysis: • Science mapping This concerns the mapping of literature on different levels of scale. Commonly, the structure of particular science fields (specialties) are mapped and elaborated graphical depictions of the relations between important nodes (documents, authors, journals or other types of entities) in the citation network are analyzed. Sometimes, the mapping involves the characteristics of a certain field’s literature, and may concern, for example, distribution of citations over language areas, geographical areas and subject areas. Science mapping could also involve the association between disciplines and research fields as well as the development of a science field over time. Science mapping is useful to information professionals involved in the organization of scientific information and it is also an important tool for the monitoring of scientific development. • Information Retrieval Citations are considered as useful supplements to keywords in the retrieval of relevant documents and have been used in various retrieval algorithms as well as in the development of document representations. Also, citation analytical methods have been applied to visualize overviews of document collections and have been implemented in Web-based applications. • Evaluation of scientific activity Here, citation counts are used as indicators of influence on research and citation analysis is applied as an evaluative tool by science 16 administrators for the assessment of universities, countries and other aggregates of scientific activity. • Collection Management Citation analysis has mainly been applied for the development of journal collections in libraries. Decisions regarding the acquisition, discontinuation or continuation of journals are supported by citation data. • History of Science Historical events of scientific enterprise could be traced chronologically by citation relations between central works and the relationship between discoveries is established through the linking of key documents through time. However, citation analysis has its limitations, which include the assumptions that have to be made in the analysis and also problems associated with citation data and sources, as discussed below. 1.2.1 Basic Assumptions Underlying Citation Analysis It is difficult to establish the underlying motivations and the significance for a citation, and they can probably never be fully elucidated. As such, one has to rely on some general assumptions. In Smith (1981, p.86 ff.), several assumptions concerning the significance and function of citing are elaborated, of which four of the more pertinent issues are quoted and discussed here. i. Citation of a document implies use of that document by the citing author This assumption incorporates that the author refers to the major part of documents used in the preparation of the citing work and that all referenced items were used. Whether a certain item is just quoted without further reading or to what extent the cited item is used, is hard if possible at all to decide. ii. Citation of a document (author, journal, etc.) reflects the merit (quality, significance, impact) of that document (author, journal, etc.) The underlying assumption in the use of citation counts as quality indicators is that there is a high positive correlation between the number of citations received and the quality. Arguments concerning the invalidity of citation counts as indicators of quality focus on the fact that documents can be cited for reasons irrelevant to their merit (e.g. negative citations). However, several studies have shown support for citation counts as quality indicators. The operationalization of other measures (non-bibliometric) of quality in comparison is found to be 17 problematic and Smith (ibid.) concludes that citation counts is are rough measures of quality. Also, one could have more confidence in counts of larger units than on individual counts. Cole & Cole (1973, p. 35 f.) also argued in favor of citation counts as indicators of quality. They reported that “[d]ata available indicate fiat straight citation counts are highly correlated with virtually every refined measure of quality”. They also warned about the misuse of citation counts, i.e. to interpret small differences as significant, and conclude that “[c]itation counts should not be used as fine measures of quality” as small differences should not be interpreted as significant . iii. Citations are made to the best possible works A better expression is perhaps the citation of “the most relevant works” in relation to the topic treated by the citing author. However, this assumption may sometimes be wrong as it has been shown that accessibility may be an important factor in the selection of references (Soper, See Smith, 1981) meaning that what is found may not always be the most relevant item. Accessibility, according to Smith, may be a function of form, place of origin, age and language and “it may be that anything that enhances the researcher’s visibility is likely to increase his citation rate...” (1981). iv. All citations are equal Taken as a major premise is that there is a cognitive relationship between the citing and the cited document. However, the strength of the cognitive relations between the citing and the cited document should not all be the same. The exact nature and strength of such a relationship is hard to characterize and measure. In spite of this, all references of a document are commonly considered to have the same status when used in citation analysis. Note though that the assumptions are not of equal importance to the different types of citation studies and this needs to be further elaborated. With regard to (i), the use of (the major part) a document is basic both for cocitation analysis and bibliographic coupling as a cosmetic referencing may not reflect the cognitive association between the citing and the cited document, bibliographically coupled documents or between the cocited documents in a valid way. Point (ii) should be essential for cocitation analysis as high citation counts of cited documents are considered to identify documents as concept markers and are applied as a prime selection criterion for cited documents to be included in the analyses. With regard to analyses of bibliographically coupled articles, point (ii) is of lesser significance as primarily the similarity between reference lists of two coupled articles are considered, not the citation impact of references. 18 Point (iii) is relevant to both bibliographic coupling and cocitation analysis. This is so, as less attentive or random referencing may lead to the absence of identified cognitive associations between citing documents treating the same topic in the case of bibliographic coupling (cf. Braam, Moed & van Raan, 1991) and in the case of cocitation analysis, less relevant associations between cocited documents would arise. Lastly, point (iv) points to a problem that should be common to both cocitation analysis and bibliographic coupling. As for now, no practicable method exists for discerning the more important associations between cocited works in a reference list, neither is there a method for the decision of which references common to bibliographically coupled articles that are the more important ones (cf. Martyn, 1964). 1.2.2 Problems of Citation Data and Sources Objections against the use of citation data in different kinds of studies might have their point of departure in the violation of assumptions used, but there also exist objections that concern the sources themselves, both with respect to citation data and to the ISI citation indexes. With reference to Smith (1981) and Vinkler (1986), thirteen problems concerning sources are mentioned in Egghe & Rousseau (1990, p. 217 ff.). Those problems that are of importance to the application of the proposed method are quoted and commented here. i. Errors This refers to errors such as misspelling, incorrect page numbers etc., due to author mistakes and transcription errors. “Whether such problems would cause appreciable error is not known, but probably they would not since there is no reason to suspect that they are systematic” (MacRoberts & MacRoberts, 1989). Systematic errors, on the other hand, could cause problems such as underestimation of citations, for example, preprints can only be indexed under “in press” or “unpublished”. ii. Synonyms This problem is foremost associated with the way the author’s name is being cited. The problem may arise under the following circumstances: authors have the same surname but different initials; a woman author may be cited in her maiden and married names; different transliterations of non Anglo-Saxon names; and misspellings. Also, variations of the abbreviated title of journal names in the reference lists of bibliographic descriptions of the citation indexes are common. 19 iii. The incompleteness of the ISI databases As the ISI method of obtaining comprehensive coverage of the literature is based on Bradford’s law, which states that only a small percentage of journals account for a large percentage of the significant articles in any given field of science, a consequence is that most journals and articles are not included. Though the body of important research in any field might be well covered, the ISI data might not fulfill the needs of local studies. iv. The dominance of English as a scientific language It is clearly so that the English language dominates the scientific communication in the Western world. A consequence is that scientific articles published in English are preferred for citations. v. The American bias The citation indexes are known to be biased towards publications from the USA. With regard to points (i) and (ii), technically, the whole text string identifying a cited reference in a bibliographic record is compared with every other such string in all other bibliographic records representing a population under study. Hence, when two text strings refer to same reference but are not completely identical, such a unit of bibliographical coupling will be omitted, if not standardized to one form. With regard to relatively small populations of source articles, semi-automatic routines may be applied for standardization purposes, increasing the number of bibliographic coupling units (Persson, 1994). Points (iii) to (v) are of no immediate importance for the evaluation of the proposed method. However, when comprehensive and exhaustive mappings are aimed at, claims of coverage of a field of research may be less valid if a considerable amount of published research is omitted on grounds of incomplete coverage, geographical or language biases. 20 1.2 3 Citation Based Science Mapping The data on which mapping and the generation of maps are based on is commonly derived from bibliographical citation databases where research articles are indexed and made accessible as bibliographic records. A bibliographic record is a representation of a research article, and contains less information than the item it represents. The information contained in a bibliographic record usually tells us who authored the article, where and when it was published as well as its subject content as indicated by abstract, title, journal title, classification codes, author key-words and assigned descriptor terms.6 The type of bibliographic records used in this study not only provides the aforesaid information but also contain references which link to the previous research that is referred to in research articles. A reference is given to a work cited in an article and is counted only once, as it occurs in the reference list of the article. One way to distinguish between references and citations is that references in a document is a property of the same, while the citation of a document informs us about the extent to which it is noticed by subsequent researchers. This is of some importance as one sometimes maps the cited works and sometimes the citing works. 6 Several terms denoting scientific, published works are incorporated in the bibliometric jargon, namely, article, document and publication. When referring to original texts, their authors’ choices of terms will be applied. The term “document” covers for other document types besides journal articles, and is applied when motivated, otherwise, the term “article” is applied. It is to be noted that though the citation databases of ISI only index journal articles (the citing items), the articles contain references (the cited items) directed to any document type. Though bibliographic descriptions of journal articles (bibliographic records) are used as input data in computational operations and calculations, rather than articles, conceptually, journal articles are analysed and are referred to also when bibliographic records are treated in practice. In document based bibliometric mapping, citation based measures of the association between documents are applied. There are three forms of citation associations between documents as follows: i. direct citations; ii. cocitations; and iii. bibliographic couplings. Direct citations means that a document is cited in another document and the strength of the association between two documents is either 0 or 1. An association of cocitation between two documents means that both documents are cited together in other documents, hence, the association is generated extrinsic to the associated documents. The strength of association between a pair of cocited documents is l...n, depending on the number of times they have been cited together. A bibliographic coupling between two documents means that both documents cite the same third document. The association between two bibliographically coupled documents is intrinsic to the documents and the strength of association is l...n, depending on the number of common references. Generally, the association (coupling) between two 21 documents is referred to as a link. A graphical illustration of the three types of citation associations are given in Figure 2-1. Figure 2-1: The Citation Associations Between Three Documents Time A dl d3 d2 The three documents in Figure 2-1, i.e., dl, d2 and d3, are published at different points in time. All three documents are associated through direct citations. Two types of document pairs are formed from them. The first pair (dl - d2) is generated through citations from d3 (cocitation). The second pair (d2 - d3) is generated through their common referencing of dl (bibliographic coupling). As the vocabulary of bibliometric mapping research is partly confusing, the separation between the concepts of measure and method are seldom clearly reflected by the use of the terms. The terms bibliographic coupling and cocitation denote measures of document association. When applying these measures, one arrives at values of bibliographic coupling strength and cocitation frequency. In the literature, the term cocitation analysis usually denotes method applications where cocitation relations are analyzed, mostly for science mapping purposes. The strength of association generated by either bibliographic coupling or cocitation is to be considered as the perceived similarity or distance between two documents where the strength of similarity is inversely related to the distance, i.e. a short distance corresponds to a high similarity and vice versa. A variety of statistical mapping techniques can be applied where input data is the values of cocitation frequency or bibliographic coupling strength, or normalized values of the same. The result is commonly a categorization of 22 documents where documents sharing a common research focus are gathered in clusters. The general definition of a cluster is a group of objects. However, in this study, the term “cluster” mostly refers to the partition of a set of research articles into subsets by means of some cluster analytical method (see Sub-section 1.4 in this chapter). The size of a subset can vary between 1 and n and a subset containing only one element is named singleton cluster. Also, the concept of cluster relevance needs some clarification. Generally, relevance is about how pertinent or connected certain information is to a given matter. When the relevance of a cluster is assessed, this concerns how well the cluster represents a coherent research theme, and different variables are applied for the measurement and assessment of relevance (see Sub-section 2.2 in Chapter 4). Other methods than cluster analytical (applying the same kind of data) may project cognitive associations between objects in a two or three-dimensional display, so that the distance between points in the projection represents the similarity between the objects. Such a method is called multidimensional scaling (MDS). A more detailed elaboration of MDS is given under Sub­ section 2.5 in Chapter 4. 1.3 Mathematical Concepts and Definitions The understanding of citation associations may be enhanced by applying concepts that are applicable to networks in general. Graph theory supplies such concepts. As such, in this study, different sets of bibliographically coupled documents (e.g. clusters) will be considered as networks which may be depicted as graphs. An undirected graph G, is constituted by a set V of vertices and a set E of edges such that each edge e e E is associated with an unordered pair of vertices.7 The existence of an unique edge e associated with the vertices v and w, implies the existence of an edge e associated with the vertices w and v and this is written as e = (v, w) ore = (w, v) (Johnsonbaugh, 1997, p. 306). In Figure 2-2, is an example of an undirected graph G. It consists of the set V = {a, b, c, d} of vertices and the set E = {ci, e^..., e$} of edges. 7 The terms used in relation to graphs, namely, “vertice”, and “edge” correspond to documents and the bibliographic coupling between two documents respectively. In more general discussions concerning clusters and their associations through bibliographic coupling, the corresponding terms “articles” and “links” are used. 23 Figure 2-2: The Undirected Graph G a c A graph G' whose vertices and edges form subsets of the vertices and graph edges of a graph G, is a subgraph of G, and G is said to be a super graph of G'. A complete graph is a graph in which each pair of vertices is connected by an edge. In Figure 2-3, subsets of Gand E constitute the subgraph G', which also is a complete graph. Figure 2-3: The Subgraph G' of the Undirected Graph G b a c An undirected graph can be presented by a symmetrical matrix. A matrix M, is a rectangular array of numbers, where M has m rows and n columns and the size of M is m x n. The numbers pertain to the elements of V and they are represented by the letters i and / and it is assumed that i and j run from 1 to n. The number connecting i with / is represented by my. A square matrix is one where the number of rows and co lumns are equal, n x n, and a symmetrical matrix is a square matrix where my = my. Hence, the associations between the elements of V can be represented. The columns and rows are labeled with the elements in V and my is equal to 1 if there is an edge between the vertices of the elements in V and 0 when there is no edge between the vertices of the elements in F (see Table 2-1). 24 Table 2-1: The Undirected Graph G Represented by a Symmetrical Matrix Note: The diagonal elements indicate the associations between i and i which are of no importance in this case. Only half the matrix is needed, (below or above the diagonal) as m¡j = W/7- When analyzing graphs and matrices, it is necessary to know some counting methods. The first is the multiplication principle which states that if an activity can be constructed in t successive steps and step 1 can be done in n\ ways; step 2 in «2 ways and step t in n{ ways, then the number of different ways is ni • »2 ’ ’ ’ wt. The second principle is permutation, which is related to the order of objects. In concordance with the principle of multiplication, the first object can (for example) be selected in four ways, the second in n - 1 ways, the third in n - 2 ways and so on. Hence, there are n(n - l)(n - 2)- • -2-1 = n\ permutations of n objects (ibid. p. 210). An r permutation of n distinct elements X\... xn is an ordering of an r- element subset of {xi... xn}. The number of r-permutations of a set of distinct elements is denoted by Pin. r) and P(n, r) = n(n - l)(n - 2) • • ■(« - r + 1). When one selects objects without regard to order, it is a combination. An r combination of n distinct elements X|... x„is an unordered selection of an r-element subset of {xj... x„}. The number of r-combinations of a set of n distinct elements is f n' denoted by C(n, r) or VJ and = + (ibid. p. 211-213). (2.1) r! r! (n-r)!r! 1.4 Classification and Cluster Analysis The second component of the two constituting the proposed method for science mapping is the method of partition. The idea of mapping science on the basis of published research articles implies a method of partition where objects are grouped to produce a classification. A classification should then fulfill the following conditions: i. it should be exhaustive; and ii. classes should be mutually exclusive. 25 This means that each object should belong to exactly one class. The forming of classes should also imply that classified objects are more similar to other objects in the same class than to objects in another class. The objective of finding such classes connects with the purpose of a set of statistical techniques with the generic name “cluster analysis”. Hence, cluster analysis involves techniques that produce classifications from data that are initially unclassified. From another point of view, cluster analysis is essentially about discovering groups in data (Everitt, Landau & Leese, 2001, p. 6). Cluster analysis is highly empirical and different methods can lead to different groupings, both in number and in content. This happens because the choice of cluster algorithm imposes a structure and cluster methods might detect clusters that have no correspondence to the real world. It is usually difficult to judge if the results make sense in the context of the problem being studied (ibid.). This concerns the fact that there are many cluster algorithms but no generally accepted best method and there is usually a subjective component in the assessment of the results. The task is, therefore, to select the most appropriate method in relation to data and empirical experiences. 1.4 1 Cluster Analytical Methods The commonly used methods fall into the following two general categories: i. non-hierarchical; and ii. hierarchical. The non-hierarchical approach requires that some objects be selected as cluster seed points around which clusters are then built. This is accomplished by assigning every object in the population to its closest cluster seed object. After this step, clusters may be split, and clusters close to one another may be combined. That is, objects are allowed to move in and out of groups at different stages of the analysis. This approach has some disadvantages according to Johnson (1998, p. 323). They include: i. it requires one to initially guess the number of clusters that is going to exist; ii. it is greatly influenced by the choice of the initial cluster seed objects. By letting the statistical program choose the seeds, the selection often depends on the order in which the data are read into the computer. As such, two researchers could perform a cluster analysis on the same set of data and produce entirely different clusters; and iii. the procedure is often not feasible computationally because there are just too many possible choices in terms of number of clusters and number of locations of the clusters seeds. In bibliometric mapping, the numbers of clusters are usually not known beforehand, which makes non-hierarchical cluster methods less applicable. 26 In general, the most widely used cluster methods are the hierarchical ones. In hierarchical methods, groups are formed by a process of agglomeration or division. The agglomeration process starts with all objects being alone in groups of one, that is, each object is considered a cluster (a singleton cluster). Objects are then gradually merged according to some algorithm until finally all individuals are in one group. The process of division begins with all objects being in one group. This is then split into two groups; the two groups are then split, and so on until all objects are in groups of their own. The general procedure of hierarchical agglomerative methods starts with the compilation of a matrix of proximity values showing similarity or dissimilarity. For example, let M be an N-N squared proximity matrix and let N clusters contain one object each and the clusters denoted 1 to N. Next, apply a scheme of agglomeration where all objects begin alone in groups of size one and groups that are “close'’ (similar) together are fused according to the steps presented below:8 8 Adapted from SPSS technical papers: Clustering Methods/ general procedure) i. Find the most similar pair of clusters i and/. Denote this similarity My. ii. Reduce the number of clusters by one through the fusion of clusters i and j. Name the new cluster p (=/) and update the matrix according to the revised proximity between cluster p and all other clusters. iii. Repeat steps (i) and (ii) until all objects are in one cluster. The result of the cluster process can be visualized by a dendrogram. A dendrogram is a two-dimensional tree-diagram which illustrates the fusions of clusters at different levels of distance at each stage of the analysis. The nodes in the dendrogram (the point where two lines meet) represent clusters and similar clusters are joined by links whose position in the diagram is determined by the level of similarity between them. An example of a dendrogram is given in Figure 2-4. 27 Figure 2-4: Example of a Dendrogram Note: This is a fusion of 10 objects, A to J. The dendrogram rescales the actual distances to numbers between 0 and 25, preserving the ratio of the distances between steps. The closest objects are A and B, which are merged with C in the next step, etc. When selecting an appropriate method of clustering, experience achieved and recorded by researchers within the field could to some extent be used as a guide. Originally, when the cocitation clustering method was developed by Henry Small and colleagues at the ISI (Small, 1973; Small & Griffith, 1974; Griffith, Small, Stonehill & Dey, 1974; Small & Sweeney, 1985), the single link cluster method was applied. The defining feature of this method is that the distance between groups is defined as that of the closest pair of individuals. Single link clustering, is known to produce straggling and loosely bound clusters, especially in large data sets, and this problem might show up as a less clear structure due to this “chaining” phenomenon. Still, single link applications seem to have been successfully used by many researchers in the context of document cocitation analysis (e.g. Small & Griffith, 1974; Griffith, Small, Stonehill & Dey, 1974; Small & Griffith, 1983; Small & Sweeney, 1985; Braam, Moed & van Raan, 1991) and a variant was used by Persson performing an author cocitation analysis9 (1994). The single link method is easy to implement and use especially when large amounts of data is to be clustered. However, as the development of cocitation cluster analysis method has attracted criticism (Leydesdorff, 1987), the use of Ward’s method has been suggested as an alternative. Ward’s method has also been mentioned as appropriate in the context of author cocitation analysis (McCain, 1990) as has the complete link cluster method (McCain, 1990; White & McCain, 1998). Ward’s method differs radically from the single link and complete link cluster methods as distances between objects are defined rather than differences between clusters and at each stage of the clustering process, the objective is to minimize the increase in the total within-cluster error sum of squares. This 9 Author cocitation analysis is a special case of cocitation analysis where the analyzed units are the authors’ names in referenced works. In author cocitation analysis, the collected research of an author is represented by the author's name. 28 method is usually used with a distance matrix of proximity data and a matrix of squared Euclidean distances is required as this method assumes that objects can be represented in Euclidean space (Everitt, Landau & Leese, 2001. p. 62).10 10 The need for using Euclidean distances motivated the exclusion of this cluster method as a candidate method for partition. See Sub-section 1.1 in Chapter 4 concerning proximity measures. Comparing complete link clustering with single link clustering, the difference is how the distance between an existing cluster and a candidate object for fusion with that cluster is defined. In complete link clustering, the largest distance between the candidate object and any object of the existing cluster is sought. This means that any candidate must be within a certain level of similarity to all members of that cluster. As mentioned, in single link clustering the shortest distance between clusters is sought. Hence, single link clustering and complete link clustering could be seen as each other’s opposites. In addition to these methods, the between groups average link appears as an alternative in this study. For this method, the distance between two clusters is the average of the distance between all pairs of individuals that are made up by one individual from each group. It was developed as an “antidote” to the extremes of both single and complete link (Aldenderfer & Blashfield, 1984, p. 40). In theory, it is possible to make some general assumptions concerning clusters generated by these methods according to differences between their algorithms. Hence, the single link cluster method would generate more loosely bound clusters whereas the complete link cluster method would produce compact clusters and the group average link method something in-between. 1.4.2 Motives for the Choice of the Complete Link Cluster Method If one can assume that the similarity between document A and document B and the similarity between document B and document C generally implies a similarity between document A and document C, a method of clustering with less severe conditions to fulfill may be appropriate (e.g. the single link method). This assumption, however, should be considered unconfirmed as previous research in cocitation clustering has shown that the chaining effect of the single link method has caused some undesirable effects like large subject inconsistent clusters when applied for cocitation cluster analysis (see Sub­ section 2.2 in this chapter). Though cocitation and bibliographic coupling are not the same, they are similar in the sense that they are both based on consensual referencing. Thus, it does not seem too far-fetched to assume that similar drawbacks might occur when the single link method is applied on bibliographic coupling data. It should also be noted that reciprocal associations between documents in a cluster through bibliographic coupling do not necessarily imply that one single cited reference is common to all the documents. Moreover, as discussed, the significance of the association between two documents through a common reference (a bibliographic coupling unit) is hard to establish. All this speaks in favor of not introducing further uncertainty on an additional level. Therefore, a method that ensures that all objects in a cluster are within a set maximal distance to each other would be preferred in order to secure coherent clusters. 29 From a graph theoretical viewpoint, such groups could be considered complete undirected graphs. Such graphs would always have a maximal degree of interconnectedness, i.e. a maximal density’ (D), where D is defined as: 2-(#£(G) N(N-V) ’ (2.2) where #L(G)= the number of edges connecting two vertices; and N= the number of vertices (Otte & Rousseau, 2002). The interval is [0, 1] and the maximum value is reached when the value of #L(G) equals the value of 7V(7V-l)/2. In this context this means that the maximum value is reached when all possible document pairs in a cluster are bibliographically coupled. Applying the complete link cluster method, one will arrive at clusters with a maximal density Z), as each cluster member is associated with every other member in a cluster, given that fusions of clusters at a level of zero association are prohibited.11 As the maximal interconnectedness is given by the method, only the strength of association (the distance) between documents varies. A maximal allowed distance between documents in clusters could be set as a way of avoiding more random associations between articles, and secure a high degree of similarity between articles in clusters. It was hence presumed that the complete link cluster method would generate coherent clusters and, therefore, also more subject consistent (relevant) clusters. However, one could always argue that this cluster algorithm may lead to a low interception of documents as one could imagine something like nearly complete graphs. " Applying the “furthest neighbour” cluster algorithm in SPSS 11.5, the fusion process continues until all objects belong to one cluster. 30 2. PREVIOUS RESEARCH 2.1 Bibliographic Coupling Bibliographic coupling was introduced by Kessler to the scientific society through a number of reports and research articles in the ’60s.12 12 The suggestion to group scientific documents on the basis of use rather than content was suggested independently by Fano (1956) and by Kessler (1958). Bibliographic coupling was primarily described as a method for grouping technical and scientific documents, facilitating scientific information provision and document retrieval. In one of the early reports, a general outline of the context in which an indexing method, concerned with countable indicators based on references, might operate was given (1960). In a subsequent report, the definition of bibliographic coupling was stated: “[a] single item of reference shared by two documents is defined as a unit of coupling between them” (1962). Based on this unit, two graded criteria of coupling were defined (ibid.): Criterion A - A number of articles constitute a related group GA if each member of the group has at least one reference (one coupling unit) in common with a given test article, Po. The coupling strength between Po and any member of GA is measured by the number of coupling units between them. G"a is that portion of GA that is linked to Po through n coupling units. (According to this criterion, there need not be any coupling between the members of GA, only between them and Po) Criterion B - A number of articles constitute a related group GB if each member of the group has at least one coupling unit with every other member of the group. The coupling strength of GB is measured by the number of coupling units between its members. Criterion B differs from criterion A in that it forms a closed structure of interrelated articles, whereas criterion A forms an open structure of articles related to a test article. The problems concerning scientific and technical information processes related to the invention of bibliographic coupling emanated from the accelerating growth of scientific and technical activities all over the world. Elaborating on the units of the scientific message, Kessler addressed the problem and need of a refined bunching process that would generate a more differentiated and individualized set of articles which would fit the need of individual researchers and groups (1960). From his point of view, the problem of indexing and the subsequent matching of print material with the scientist’s need, was the main problem. In the context of a proposed science communication system, he stated that “[t]he goal is to discover certain measurable or countable indicators that reflect on the operational background of a scientist in terms of the four components previously mentioned” (ibid.). The four components that he referred to were the Man, the System, the Operation and the Results, each referring to the formal pattern common to the scientific work. According to him, these four components should be used to index documents. He also suggested that a scientific document is a reflection 31 of the operational history prior to publication and that a scientist’s information needs are also determined by his operational background. Therefore, the best way to determine a man’s need for information is to examine him in terms of the same four components that were used to index the document. According to him, the component Man would be sufficiently described by the cultural environment that best describes the scientist’s position in the intellectual community and the reflection of this environment may be found in the bibliography of citations that he finds necessary to append to his remarks. In other words, Kessler suggested that references could serve as countable indicators of the intellectual environment in which scientists find information useful for their current work. One could then assume that a shared intellectual environment (as reflected by common references) between two documents might indicate a relationship and this information might be used to facilitate information provision. In a subsequent report dated 1962, Kessler applied bibliographic coupling to a test population of 40 documents from the field of radio engineering in order to test if a number of scientific documents bear meaningful relations to one another. He found that bibliographic coupling was able to partition this population into valid, related sub-groups. In order to be able to make any generalizations at all as to larger populations and other fields, he subsequently carried out an experiment where the automatic processing of a population of scientific documents (36 volumes of Physical Review) re sulted in a grouping of 8,521 articles in concordance with criterion A (1963a). This experiment had its starting point in the population of 8,521 articles, where common references between each member of the population were sought. Kessler (ibid.) reported the outcome from processing one volume (265 articles) and claimed that this experiment proved the existence of groups of documents Ga(P0) related to one another through the coupling with a fixed document Po in a well-established field of science. This process could then be iterated. The outline of application areas of this method concerned indexing, classification and information retrieval. He intellectually and by reference connects this document with the previous report from 1960 by pointing out the significance of this method to information retrieval, i.e., once a particular document is identified as relevant, a retrieval system could also retrieve Ga(Po). He also highlighted some properties of the method: • The method is independent of words and language as all the processing is done in terms of numbers. • No expert judgment is required - the text is in fact abundant. • The group of documents associated with a fixed document, GA(P0), extends into the past as well as the future. • The size and growth of a Ga will reflect the continuous impact of a fixed document and the groupings will undergo changes that reflect current usage and interests of the scientific community. 32 • Documents similar in the way that they all share references with a third document could be seen as this document’s “logical references”, which could be substituted for its real references. Conclusively, Kessler showed in this paper the existence of subject relatedness between bibliographically coupled documents. In the same year, Kessler published another paper (1963b) where he further elaborated the application of bibliographic coupling in the context of information retrieval by trying to establish a factual background that could guide the design of an experimental science communication system. Bibliographic coupling was applied to a population of 8,186 articles from the Physical Review and reported as ten case histories, each illustrating an information retrieval problem. Different strategies of bibliographic coupling were applied, where the effects of enlarging or diminishing the search span by assigning Pos serially first (in the case of enlarging) or last (in the case of diminishing) in the list of the available literature were tested. Kessler concluded that bibliographic coupling can be applied to a large body of literature and that the process operates both in the future and in the past, relative to the position of Po. This showed that bibliographic coupling could be used to identify the life span of a given literature. In 1965, Kessler, still using data from the Physical Review for his experiments, compared groups formed according to the Analytic Subject Index and by bibliographic coupling. The aim of this experiment was to investigate how bibliographic coupling compare with results obtained by standard methods. He concluded that there was a high correlation between groups formed by bibliographic coupling and groups formed by analytical subject indexing. However, he pointed out that the report did not pass judgment on the utility of either method to any specific application. In a review article, Weinberg (1974) covered the major part written on bibliographic coupling up to the publication of her article. She concluded that “[a]t this point, bibliographic coupling does seem to be a useful tool for studying the ‘science of science’ - citation patterns, the useful life of literature, most cited journals etc”. However, she reflected on how citation behavior affects the standardization of the citation “unit” and put forward the meaning that the notion of “meaningful groups”, claimed by those who advocates bibliographic coupling, may well constitute a problem. What she was trying to say was that since Kessler’s experiments were done on documents in one field, there was already a meaningful group to begin with. Therefore, only a test on the scale of SCI would show if bibliographic coupling would work well in a complex and interdisciplinary environment. However, the first attempt to test the validity and effectiveness of the bibliographic coupling technique for detecting subject relatedness between documents on a more heterogeneous population of documents and on a large scale, was not performed until more than twenty years after Kessler’s 1963 reports. One reason for this was probably the technical restrictions imposed by 33 existing (at that time) computational resources and the problems in accessing large amounts of citation data. In 1984, Vladutz and Cook carried out an experiment with 10,000 randomly selected documents from the SCI which served as test documents for which bibliographically coupled publications from the entire 1981 database were sought. The large data file covering a multitude of scientific disciplines used in this experiment, corresponded well with Weinberg’s claim in 1974 of an interdisciplinary environment as a prerequisite for the evaluation of bibliographic coupling. The questions to be answered with respect to this experiment concerned the frequency of bibliographic coupling links within the file and the degree to which these links are meaningful. It was found that 90 percent of the input articles that have references, yielded a group of at least two coupled items. Looking back at Kessler’s experiments in the ’60s, Vladutz and Cook wanted to test more extensively the hypothesis that strong bibliographic coupling links imply strong subject relatedness. The evaluation of subject relatedness was performed by small groups of experts with a scientific background and trained in assigning brief subject descriptions to groups of documents generated by cocitation clustering. Lists of 300 randomly selected test documents together with their strongest coupled articles were presented to the experts. It was found that in over 85 percent of the cases, the articles proved to be closely related by subject to the test documents. Vladutz and Cook concluded that the utilization of bibliographic coupling in a very large citation database was practically feasible and that valid results as to subject relatedness were achieved. The hypothesis stated in this research was that bibliographical coupling “[m]ay prove to be the easiest approximation to an algorithm for revealing the semantically closest neighbors of publications”. A year before Vladutz and Cook published their results. Sen and Gan (1983) had published a purely theoretical article on bibliographic coupling. Their point of departure was a statement by Martyn in 1964 where he argues “[t]hat bibliographic coupling is not a unit but merely an indication of the existence of the probability, value unknown, of relationship between two documents”.13 The two researchers felt that in spite of the attention that previous works on bibliographic coupling had attracted, the method had hardly been taken seriously and that there was a need for a theoretical elaboration. With a point of departure in an M x N hypothetical Boolean matrix, where elements indicated a citation relationship between rows (citing documents) and columns (cited documents), the grouping of coupled documents in bibliographic cliques and clusters was elaborated. The notion “clique” is here equivalent to Kessler’s grouping principle GB, and “clusters would be formed by the populations which have at least one member having coupling with another member whereas no member of one cluster will have coupling with any member of another separate cluster”. 13 The meaning of this statement in short is that the fact that two documents have a reference in common is no guarantee that both documents are referring to the same piece of information in the cited document. Hence, bibliographic coupling is only an indication of the existence of the probability of relationship between two documents. 34 With regard to the central issue of cognitive resemblance between bibliographically coupled documents, a measure of coupling strength, the Coupling Angle (C.A.) was suggested. The Coupling Angle was expressed as: (¿W ‘ DOj\Dok • Dok) (2.3) C.A. is the coupling angle for citing documents j and k. Doj and Dok are the binary vectors of document / respectively k. The coupling angle C.A. is a geometric interpretation where the C.A. takes the maximum value of 1 if two Boolean vectors are parallel and 0 if they are rectangular. Two documents may be considered to be concerned with a related topic if the angle between vectors representing documents does not exceed a given angle 0 (0° < 0 < 90°) (Glänzel & Czerwon, 1996). Lacking a theoretical basis as well as empirical evidence for the determination of a threshold of coupling strength, Sen and Gan suggested a semi-arbitrary approach with a cut off value of 0.5, which corresponds to 0 = 60°. The question of cognitive resemblance related to bibliographic coupling was also pursued in Peters, Braam and van Raan (1995). These researchers tried to find out whether relatively strong cognitive resemblance within groups of documents, bibliographically coupled by one and the same highly cited item, is present in an interdisciplinary field, i.e., Chemical Engineering. This was operationalized by measuring word-profile similarities between the citing documents. It was found that word profile similarity within groups sharing a citation to a highly cited publication was significantly higher than between documents without such a relationship. Hence, such cognitive resemblance was found to exist, supporting the claim that these bibliographically coupled documents did represent work of the same research specialty. In Glänzel and Czerwon (1995 and 1996), it was shown that bibliographic coupling can be used to identify “hot” research topics as represented by so called “core documents”, which were identified through the application of appropriate thresholds for both the number of common references as well as the strength of coupling links. Using the whole annual accumulation of the 1992 volume of SCI, about one percent of all documents was found to be core documents. 14 A detailed analysis of both key words in titles and indexing terms indicated the representation of important research front topics and through several expert questionings it was found that most core documents belonged to a few high impact documents of a specialty. Performing a cross­ national citation analysis covering two years, Glänzel and Czerwon (1996) found further empirical evidence substantiating the claim of high impact of core documents: (1) only a small share (15.7%) of core documents were not 14 The data comprised 511,899 articles, notes and reviews, and only the document type “letters to the editor” was excluded on grounds of generally not belonging to research fronts. 35 cited, (2) almost two thirds of core documents were cited above the average and (3) a relatively large share (10%) of core documents were highly cited.15 15 For the assessment of the impact of core documents published during 1992, citations received during the period 1992-1993 were compared with corresponding journals’ citation impacts. “Highly cited” meant that a core document had received at least five times as many citations as the average article in the journal in which it was published. The method presented proceeds from the model suggested by Sen and Gan (1983) and uses the C.A. as a measure of the coupling strength. Glänzel and Czerwon restricted their analysis to a subset of coupled documents where each document was coupled with at least ten coupling links with a minimum C.A. of 0.25 to other documents. The choice of thresholds was based on both theoretical considerations and empirical findings. According to the researchers, a lesser number of coupling links could bring about that documents published in series might influence results, whereas a greater number of coupling links would eliminate smaller research topics. They also claimed that a certain filtering of noise is necessary in order to avoid less characteristic coupling links between documents and that a value of the C.A. considerably lower than the stipulated would not accommodate this need. Also, too high a value of the C.A. would dramatically diminish the number of coupling links, leading to a serious decrease of found documents. The researchers concluded that documents connected by strong bibliographic coupling links can provide insights into the structure of research fronts and be applied for science mapping purposes. They also highlighted that bibliographic coupling has several advantages in comparison with cocitation clustering, the most important being the possibility to capture the early stages of a specialty’s evolution. 2.2 Cocitation Analysis The cocitation frequency, a measure related to bibliographic coupling, was independently introduced in 1973 by Small and Marshakova. This form of document coupling was defined as the frequency with which two documents are cited together (Small, 1973). The cocitation strength is then defined as the number of identical citing items. Small (ibid.) also gave a formal definition of cocitation coupling as follows: If A is the set of documents which cites document a and B is the set which cites b, then AnB is the set which sites both a and b. The number of elements in AnB, that is n(AnB), is the cocitation frequency. The relative cocitation frequency could be defined as n(A n B) n(A U B). When measuring cocitation strength, the degree of association between documents as perceived by the population of citing authors is measured. Hence, to be strongly co-cited, a large number of authors must cite two earlier works. Small argued that due to the dependence on authors, cocitation patterns can change over time, just as vocabulary co-occurrences can change over time as subject fields evolve. Furthermore, Small noted that bibliographic coupling 36 is a fixed and permanent relationship because it depends on references contained in coupled documents. That is, once two documents are published, their coupling is established through their references, whereas the cocitation strength between any two documents will vary over time. Small argued that frequently cited documents could be assumed to represent key concepts, methods or experiments in a field and that cocitation coupling could indicate relationships between such documents. To illustrate his ideas, Small empirically tested how cocitation patterns would develop in a set of highly cited particle physics documents from the first quarter of the 1971 SCI. Setting a threshold of nine citations, all references in documents citing a highly cited fixed document were collected in one cluster and the cocitation relations in this cluster were analyzed. Ten important documents finally constituted a cocitation network which was analyzed on a detailed level both as to different citation relations as well as to content. Small found no clear relationship between bibliographic coupling strength and cocitation frequency, but direct citations and cocitation seemed to correlate. In several instances, highly co-cited documents were not bibliographically coupled at all. Small suggested that these results indicated that bibliographic coupling should be a less reliable indication of subject relatedness than cocitation coupling. Small concluded “[t]hat an interpretation of the significance of strong cocitation links must rely both on the notion of subject similarity and on the association or co-occurrence of ideas”. Small assumed the usefulness of the following two information retrieval applications based on cocitation couplings: i. a secondary index based on highly co-cited documents which would allow sequential searches through a citation index; and ii. the creation of a cluster or core of earlier literature for a particular specialty, serving as a basis of an SDI system .16 16 SDI (Selective Dissemination of Information) is a current awareness service providing researchers with current and pertinent publications on a specified research topic. Small also foresaw the use of cocitation in the study of the specialty structure of science and as a way to monitor the development of scientific fields and the assessment of interrelationships within and between specialties. In another article in the following year, Small and Griffith (1974) reported an experiment where a computer based system was used to identify clusters of highly interactive documents in science. This experiment aimed at a technique that would make it possible to explore the entire structure of specialties and their relations. The authors argued that an overall view of discipline structure could be obtained by analyzing citation patterns between journals, but for the purpose of revealing the fine structure of science, this was too broad a unit. Partly based on evidence that the onset of rapid specialty growth is accompanied by the emergence of key documents which are quickly and 37 frequently cited, the researchers’ hypothesis was that the specialty structure could be revealed by clustering frequently cited documents. In this investigation, three percent of one quarter of the annual 1972 SCI file was applied, and those specialties, which were most active during the early part of 1972, were identified. Using the single link cluster method, clusters were formed at different levels of cocitation strength and presented as graphs.17 A central question was whether clusters correspond to identifiable subject matter specialties. Small and Griffith mentioned several possible errors, namely: • specialties which are not socially or intellectually related to one another could be linked together; • the clusters themselves could be fragments of sing le specialties; and • clusters could bear little or no relationship at all to the specialties. The evaluation of their method was pursued by inspection of clusters at different levels. Known specialties like “particle physics ', “nuclear physics”, “reverse transcription” and “Australian antigen” were immediately recognized, though at different levels of coupling strength. Connections by cocitation between documents in clusters were also inspected and recognized as valid trails of history of research in some specialties. The researchers found that cocitation links between clusters at different levels generally linked to appropriate clusters. They also tried to examine word usage of titles in documents citing documents in clusters. Moving through a particular network, discontinuation of a vocabulary would indicate failure to group together documents of a certain specialty. On the other hand, identical or very similar use of vocabulary but the absence of cocitation links would indicate failure to group documents of a certain specialty. With a point of departure in these assumptions, Small and Griffith examined the two physics’ clusters by creating word profiles for each of the cited documents. Such a word profile would be constituted by the four most frequent title words from all documents citing a particular document. The researchers found that the vocabulary within the clusters was consistent with their perceived subject content. They argued that the very existence of document clusters which, by definition, have a high degree of internal linkage, should be strong evidence for the specialty hypothesis. They also stressed that specialties of science did not seem to be isolated from one another but connected by weak links as almost all documents in their sample were linked, although tenuously. Their final conclusion was that science and its literature could be conceived as a network of specialties, each specialty being the centre of an interactive and intense communication system. 7 Sub-files of cocited pairs were created at four levels, i.e., level 1, level 3, level 6 and level 10. Level 1 comprised all generated pairs of cocited documents, level 3 comprised pairs cocited at least three times, level 6 comprised pairs cocited at least 6 times and level 10 comprised pairs cocited at least 10 times. 38 In a sequel to the above article (Griffith, Small, Stonehill & Dey, 1974), Griffith, Small and their fellow researchers reported their attempts to create maps on different scales of scientific literature by applying the same set of bibliographic records. The intention was to create an overview of all highly cited documents in natural science, to reflect a single specialty in detail and to present a new clustering technique, namely, cluster cocitation. Applying this technique, weaker links of co-cited document pairs connecting any two clusters on a cocitation threshold lower than stipulated for the inclusion of co­ cited documents in clusters were used. Hence, cluster cocitation is a count of the number of times that documents in two different clusters have been cited together. A preliminary map of science, constituted by interconnected clusters was produced and it was found that a majority of clusters were small, containing three to four documents, while a few - biomedicine, chemistry and physics - were of considerable size, being the major poles. Regarding the phenomenon of small clusters, the researchers suggested that these clusters might be fragments of larger clusters that would not emerge at the level of applied thresholds. At several points, the researchers encountered macro clusters which could not be analyzed as single specialties, e.g. “cancer research - reverse transcription”. Taking this cluster as an example (similar results were reached from analyzing the remaining macro clusters), the researchers analyzed the distribution of strengths of links within and between sub-clusters, i.e. groupings within macro-clusters. They found that about half of all possible links18 between sub-clusters were absent and a negligible number of links exceeded an average of two cocitations. In addition, only a small number of sub-clusters with internal linkages of an average cocitation strength less than three cocitations were found. This led the researchers to conclude that the structure of macro clusters could be reduced to the following two components: 18 The number of possible links refers to the number of 2-combinations of a set of n distinct documents, . See equation 2.1 under Section 3 in Chapter 2. i. the internal structure within small groupings of documents; and ii. a structure of few linkages greater than zero which hold the smaller groupings together. In the above-mentioned experiments, (Small & Griffith, 1974; Griffith, Small, Stonehill & Dey, 1974) integer citation thresholds were used to select highly cited documents, and cocitation was defined on an integer basis. Also, clustering thresholds were set in terms of the integer cocitation frequency. However, these early approaches presented some difficulties which include: • very highly cited documents, such as biomedical methodology documents had to be removed in order to prevent the creation of very large macro-clusters; i.e. C(w, r) = C(n, 2) or 39 integer cocitation counts introduced a size dependency, i.e. highly cited documents also tended to be highly co-cited, which biased analyses against smaller research areas; and • it is well known that there are differences as to the length of reference lists between disciplines, e.g. between mathematics and biomedicine. When an annual slice of science is analyzed, a higher reference intensity per document could affect the cocitation clustering in the following two ways: i. it increases the number and proportion of items from the discipline that has a higher reference intensity per document; and ii. it increases the strength and density of cocitation links formed amongst documents from the discipline that has higher reference intensity per document. Consequently, in 1975, cocitation normalization was introduced to overcome some of these problems (Small & Sweeney, 1985). Applying either the Jaccard coefficient or the cosine function for normalization, one could partially overcome the problems of highly cited method documents and size dependency. The “Jaccard coefficient” (commonly referred to as the Jaccard's index) is a well-known measure of the similarity S between two objects A and B, which counts the number of common attributes divided by the number of attributes possessed by at least one of the two objects: AR B Au B (2.4) In the context of cocitation analysis, this function is expressed as: ^csv (C,+C7-Cy)' (2.5) and the cosine function (commonly referred to as Saltón’s cosine formula) is expressed as: (2.6) where: NCSy = the normalized coupling strength between document z and/; Cg = the number of cocitations of document i and/; C¡ = the number of citations of document z; and 40 Cj = the number of citations of document j. Both measures take values in the interval [0,1], Based on experiments on the 1979 volume of the SCI, Small and Sweeney (ibid.) proposed the following improvements to the cocitation cluster technique: i. fractional citation counting; and ii. variable level clustering, with a maximum cluster size limit. The first step in the cocitation cluster method is to set a threshold for the minimum number of citations a document needs to receive in order to participate in the clustering. Using fractional citation counting, each reference is assigned a weight corresponding to the length of the reference list, e.g., if a reference list has the length of ten items, then each item is assigned 1/10. This procedure generally has the effect of giving documents with short reference lists greater weight relative to documents with longer ones. Hence, some of the problems concerning the bias toward high referencing fields should be avoided using fractional counts. It was also found that fractional counting increased the range of subject matters covered by clusters. As the optimum cocitation level from the standpoint of recall and precision varies from specialty to specialty, the difficulty lies in the selection of the best cluster version. In order to deal with this problem, a strategy of variable level clustering, where clusters could be generated at different thresholds, was used. Using maximum size as a limiting parameter, a cluster would be generated at the lowest possible cocitation strength level, provided it did not exceed the specified cluster size. If so, the program would increment the cocitation level and try to form a cluster again on a higher level. This strategy breaks large clusters into smaller fragments, but since it allows the initial cocitation threshold to be set lower, it also allows for smaller clusters to become larger. The conclusion was that both fractional citation counting and variable level clustering improved the results when cluster analysis was applied on an interdisciplinary database like the SCI. The methodology developed by Small and allies have been seriously criticized by Leydesdorff both on methodological and on theoretical grounds (1987). Leydesdorff criticized the choice of methods preceding the model building and the exclusive focusing on the validation of the outcomes on behalf of the validation of methods. Leydesdorff meant that based on ad hoc hypotheses, which were basically wrong, Small and co-workers had assumed that “[t]he very existence of document clusters/.../is strong evidence for the specialty hypothesis” (Small and Griffith, 1974). Leydesdorff argued that this was a fallacious argument as cluster analysis always generates a cluster structure and that the real question was to determine what the structure represents. The methodological decisions were further criticized with regard to the choice of the single link cluster method, on grounds of not generating results consistent 41 with results obtained by other analytical techniques.19 The single link method is also known to produce loosely bound clusters and Leydesdorff suggested that results derived from the application of this method m ight well be artefacts of the method and not reflect the structure of science. 19 “Other analytical techniques’’ refers to multidimensional scaling, factor analytical approaches and Ward’s cluster method. From a research policy point of view, the concept of cocitation cluster analysis was also heavily criticized by Oberski (1988), where one of several points of criticism was directed to the statistical instability resulting from both the application of the single link cluster method and the arbitrary application of threshold settings. Oberski concluded that “[it] remains unclear how one could possibly distinguish between perhaps real effects from statistical effects” (ibid, p. 448). The claim that cocitation analysis is a useful tool to map subject-matter specialties was further examined by Braam, Moed and van Raan in 1991, which have developed a method using quantitative analysis of content-words related to articles. Unlike the basic cocitation cluster method, here the authors investigated both documents grouped by the principle of solely cociting documents of a particular cluster of co-cited documents, as well as the cocitation clusters themselves. The single link cluster method was applied for the grouping of cocited articles. Based on findings, the authors concluded that the question if all topics covered by a data set can be identified by cocitation clustering, can only partially be answered by comparing results for different sets of thresholds of (normalized) cocitation strength as some research areas might lack a consensual referencing. Still, findings suggested that cocitation clustering does display research specialties, although these may be fragmented into several clusters. It was also found that cocitation clustering only partially revealed the literature relevant to identified research topics of the citing literature (a specialty’s current work) and that interrelations between clusters seemed to correspond to cognitive relations on a higher level then research specialties. The authors concluded that the method applied provides a useful instrument for the description and evaluation of cocitation analysis in terms of the cognitive content of clusters, cluster coherence and differentiation as well as the recall of specialties current citing articles. 42 3. SUMMARY AND FOUNDATION FOR THE RESEARCH DESIGN 3.1 Origination of Methods and Direction of Development Some resemblances as well as differences are at hand when comparing bibliographic coupling with cocitation analysis. To begin with, both bibliographic coupling and cocitation analysis were originally assumed to be applied for information retrieval purposes, though cocitation analysis was additionally and originally suggested for science mapping. One can also see a partly parallel development of these methods. In both cases, the original elaborations were followed by experiments on large scales in order to be able to generalize findings, in particular to interdisciplinary contexts.20 The issue of cognitive association between documents was then further elaborated and normalized measures for the association of documents suggested. 20 Interdisciplinary contexts meant in practice that the mutltidisciplinary database SCI were applied for the experiments refered to. More precisely, a multidisciplinary environment would allow for the mapping of cogntive relations between different disciplines, i.e. the mapping of interdisciplinary researches. 21 Still, a few research articles have applied bibliographic coupling for science mapping purposes with results that indicate that it may be be successfully applied for science mapping purposes (Sharabchiev, 1988; Persson, 1994 and Jameving, 2001). However, the applicability of bibliographic coupling as a science mapping tool were not exhaustively elaborated in neither of these articles. However, the relation between bibliographic coupling and science mapping is considerably weaker in comparison with cocitation analysis. The application of cocitation analysis in science mapping and the generation of science maps, follow a clearly discernable track with a series of connected articles. This development has not been paralleled by bibliographic coupling. Based on profound empirical findings and theoretical considerations, bibliographic coupling was highlighted as a science mapping tool first in the ’90s (cf. Glänzel & Czerwon, 1995 and 1996).21 By that time, the cocitation cluster technique had undergone several adjustments and refinements, new forms of cocitation relations had been explored and a corpus of research articles providing empirical experience was already at hand. 3.2 Comparison of Properties of Methods A complementary aspect of bibliographic coupling when compared with cocitation cluster analysis is that the more current published research can be mapped. Hence, “[snapshots of early stages of a specialty’s evolution...” can be provided (Glänzel & Czerwon, 1996). This is so because there will always be a time lag between the current published research and the generation of a sufficient number of received citations that can facilitate stable sets of cocitation data for mapping. Hence, cocitation analysis has a clear shortcoming in comparison with bibliographic coupling with regard to topicality, as bibliographic coupling can capture new lines of research as soon as findings have been published. Another marked difference between cocitation analysis and bibliographic coupling concerns the identification of research specialties. According to 43 Small’s theory of cocitation analysis, research specialties are identified through the cocitation of highly cited papers. These papers are then seen as key-documents of a specialty (Small, 1974). These key-documents are regarded as symbols or markers of important concepts and the cocitation of such documents is then the measure of association or co-occurrence of ideas (Small, 1973; 1977). Hence, “cocitation identifies relationships between papers which are regarded as important by authors in the specialty, but which are not identified by such techniques as bibliographic coupling” (Small, 1974). The clusters generated on basis of the grouping of such cocited papers should then be representations of scientific specialties (ibid.). With regard to bibliographic coupling, claims can generally not be made that documents represent key-concepts or have a central meaning to researchers of the field under study. Hence, applying cluster analysis based on bibliographic coupling, one could not make the same claim of identifying the cognitive core of a research specialty, i.e., generally, there would be no selection criteria for the identification of significant articles.22 Findings concerning “core documents” may point to an exception, as these have indicated that these often are high impact papers of specialties (cf. Glänzel & Czerwon, 1996). One could, however, assume that more empirical research is needed to elucidate the relation between the criteria of core documents and citation impact. Therefore, and as for now, a perhaps more justified assumption would be that current research themes, rather than the cognitive structures of specialties, may be mapped through cluster analysis based on bibliographic coupling. Weather such research themes would reflect core issues of a specialty or more peripheral aspects may perhaps be reflected by subsequent citations. As the cocitation cluster approach has rendered some severe critique (e.g. Leydesdorff, 1987; Oberski, 1988), the question whether this method in fact mirrors the specialty structure of science may not be conclusively solved. It should however be clear that both methods, cocitation analysis and bibliographic coupling, have the ability to group cognitively related documents, hence their significance for scientific information provision is obvious. 22 However, nothing hinders that only coupling relations based on the more cited references are applied for the establishment of associations between source articles, as was done in Peters, Braam & van Raan, 1995. However, the appropriateness of this approach as a standard procedure could seriously be questioned. 3.3 Presumed General Problems of Citation Based Document Mapping It seems reasonable to assume that some general problems exist comprising both cocitation clustering and bibliographic coupling clustering. On the basis of previous findings from cocitation clustering and theoretical considerations, a number of problems including their inherent relation to one another seem evident. They are related to: i. cluster size ; ii. threshold settings; iii. fragmentation; 44 iv. coverage and of topics; and v. basic assumptions. Concerning cluster size (i), the problems encountered in cocitation analysis was for all the creation of macro-clusters, which may be considered as an effect of the applied cluster method, which consistently has been the single link method, and several adjustments of the cocitation cluster method have subsequently been considered necessary. The partition of sets of documents by cocitation clustering has commonly also led to the generation of a large number of singleton clusters and smaller sized clusters, the number of which increases by the raising of coupling threshold. As a too large number of clusters in itself would introduce noise and hinder any intelligible and comprehensive analysis of structural aspects, smaller sized clusters are usually excluded in the analysis, if at all accounted for. The issue of cluster size is related to both (ii) and (iii). The purpose of setting thresholds (ii) in cocitation clustering has aimed at the fdtering out of noise and identification of the more significant representatives (the more cited documents) of research specialties, (Small, 1973; Griffith et al, 1974). The basic difference between cocitation clustering and clustering of bibliographic coupled articles with regard to threshold settings is that in the former case, aggregations of citations guide the selection of items that should participate in the analysis. Set thresholds of cocitation strength may additionally restrict the original population of documents. In the case of bibliographic coupling, the selection of articles is based on the strength of similarity between objects, that is, articles lacking significant associations to other documents are filtered out. One may additionally apply thresholds regarding the number of links between bibliographically coupled articles at a certain minimum coupling strength (or normalized coupling strength) and in this way filter out articles that are less central in the network of interrelated articles. The setting of an “appropriate” cocitation coupling threshold for a particular specialty is, however, difficult and heuristic methods are usually applied (Small, 1977). With the exception for the strict criteria of “core documents”, the problem of coupling threshold addressed by Small likewise applies to bibliographic coupling. Related to the issues of threshold setting and cluster size is the question of fragmentation (iii). Findings from research in cocitation analysis have suggested that cocitation clustering does display research specialties, although these may be fragmented into several clusters (Braam, Moed & van Raan, 1991). An assumption could be made that the cause of fragmentation sometimes could be due to a too severe threshold setting, but also, one could 23 Applying the single link cluster method, an attempt to establish intervals of normalized coupling strength in which stable cluster sizes occur and the generation of macro clusters is avoided has been made (Braam, Moed & van Raan, 1988). However, it seems rational to separate the question of “noise and signal”, from the question of method of partition. 45 assume, sometimes reflect actual circumstances when a research area is split up in diverging directions of research and re-modeled. Another important issue for both cocitation analysis and bibliographic coupling should be the extent to which topics covered by a population of research articles are identified by the applied method (iv). With regard to cocitation clustering, research in this issue (Braam, Moed & van Raan, 1991) has shown that low “recall” of the “current work” of specialties is related to a lack of consensus as to the previous literature. 24 25Generally, for cluster analyses based on either cocitation relations or bibliographic coupling relations, when there is a lack of consensual referencing, shares of the literature on the same topic may be lost. What actually is mapped, is a slim strip of consensus that associates a fraction of all documents of a specified field under investigation (the selected document population), which subsequently is partitioned in subsets (clusters) which are claimed to represent research specialties in the case of cocitation cluster analysis. Hence, the exhaustiveness of the mapping of a research specialty is usually unknown when applying citation based mapping methods. 24 A low recall in this meaning means that only a smaller share of source articles that are semantically similar direct references to the cocitation cluster that represent a research theme that is common for these source articles. 25 This is to some extent a technical question and one could imagine that the frequency of references in a fulltext to a document could be considered. Lastly, with regard to (v), the basic assumptions should be nearly the same for both cocitation cluster analysis and the proposed method based on bibliographic coupling. In short, the cognitive association between the citing document and the cited document should be based on the use of the citing document, and the selection of cited documents should not be influenced by randomness. As current mapping techniques do not usually assign weights to a cited reference, ' the equal treatment of cited references may be considered a problem leading to a loss of precision, should it be motivated to establish the cognitive relation between a citing and a cited document. A difference between cocitation analysis and bibliographic coupling is that in bibliographic coupling analyses no obvious selection criteria of quality exist, hence, the number of basic assumptions is somewhat reduced with regard to bibliographic coupling. With regard to the different types of couplings, cocitation coupling and bibliographic coupling, the significance of a single coupling (one cocitation or one shared reference) should be related to how well appropriate basic assumptions hold, which in practice is not feasible to establish. 3.4 Methods of Partition Though much of the prerequisites necessary for the development of bibliographic coupling into a practicable mapping tool are at hand through previous research (for all the establishment of cognitive relations between coupled documents), there is a lack of empirical experience concerning principles of partition of bibliographically coupled document populations. As 46 previously described, the cocitation cluster analytical method has been criticized on several grounds of which an important one is the choice of cluster method. Undesirable effects like the chaining phenomenon may lead to less coherent clusters, especially when less homogenous populations are analyzed (interdisciplinary research settings) due to spurious links between documents. Therefore, as discussed in Sub-section 1.4.2 in this chapter, the idea of applying a cluster method that would not have this drawback is appealing. Certainly, other problems may be introduced, but the testing of a cluster method that could be expected to generate more coherent clusters seems motivated. The idea of such coherent groupings of documents was already put forward by Kessler in 1961 as “criterion B”, where an interrelated group of documents is a group where every member has at least one coupling unit with every other member of the group. Likewise, the notion of “cliques” of bibliographically coupled documents, suggest this type of grouping of interrelated documents (Sen & Gan, 1983). The criterion stated for the generation of GB groups (Kessler, 1962) and subsequently the notion of “bibliographic cliques” (Sen & Gan, 1983), are, however, insufficient from an algorithmic/practical point of view. Though the condition to fulfill for the generation of GB groups applies to the notion of complete graphs or “cliques”, this condition could be fulfilled in different ways, as illustrated in Figure 2-5, where two possibilities to form a complete subgraph where n >2 are indicated. Figure 2-5: Two Ways to Form a Complete Subgraph Note: Points denote documents and lines bibliographic coupling links. Hence, from a practical point of view, it is necessary to decide on the most appropriate algorithm. 47 3.5 A Foundation for the Research Design It can be concluded on theoretical grounds that the proposed method cannot substitute the cocitation cluster analytical method. Based on the assumption that highly cited documents represent important concepts of a specialty, claims that research specialties’ cognitive structures can be mapped are made by the advocates of the cocitation mapping approach. Such claims cannot be made with regard to the proposed method as no conclusive criteria exist that would identify the key-documents of a specialty on the basis of bibliographic coupling. It seems, however, reasonable to assume that specialties would be identified by the proposed method, but it would remain unclear to what extent core issues or more peripheral issues of a specialty would be mapped. Also, the question what really may be mapped applying citation based science mapping methods in general remains not fully elucidated, which, amongst several things, concerns the question of the exhaustiveness of citation based science mapping. In addition, with regard to both methods, problems remain with the appropriate setting of thresholds, the choice of the most appropriate cluster method and also with the relation between the two. Here, one could hope, empirical experience may by time be generated that would at least shed some light on these complex issues, but for now, more heuristic methods must be applied. In all, it could be argued that many elements of uncertainty are attached to citation based science mapping, in particular when the objective is set to map the cognitive structures of specialties. However, should the objectives be related to scientific information provision or information sharing needs, the uncertainty attached to citation based mapping in general should have lesser importance as the topicality and relevance of obtained information should be the first priority, not the exactness nor the exhaustiveness of the mirroring of specialties cognitive structures. A rational stand-point would be that more empirical evidence is needed if rightful claims of valid depictions of specialties’ structures could be made and that this may imply complex methods where several techniques are combined (cf. Braam, Moed & van Raan, 1991). On the basis of the aforementioned reasons regarding limitations of the proposed method and citation based science mapping in general, a balanced presumption of what could be mapped applying the proposed method is needed. Given is the ability of the proposed method to generate coherent cluster with regard to the applied measure of document similarity (bibliographic coupling) and the chosen cluster method. On basis of previous findings regarding subject similarity between bibliographically coupled documents, it could also be presumed that such coherent clusters generally would be subject coherent, given appropriate thresholds of coupling strength or normalized coupling strength. Applying the proposed method, the most current and consensual research would be reflected, though central aspects and outlines of specialties' cognitive structures would generally not be expected. 48 It can from this reasoning be presumed that the proposed method may serve a complementary purpose in relation to the cocitation cluster method in the context of scientific information provision. Hence, the identification of current and coherent research themes (rather than specialty structures) as mirrored by reciprocally subject related documents should be the optimal outcome expected when applying the proposed method. The desirable ability of bibliographic coupling applications to map the most current published research should be contrasted with the retrograde mappings of the cocitation approach. Bibliographic coupling methods may therefore provide scientists with more current and valuable information. In comparison with the application of bibliographic coupling for searching and retrieval of related documents, the assumed capability of the proposed method to generate subject coherent groups of documents would mean a considerable progress in comparison with the retrieval of a number of rank ordered bibliographically coupled documents as more information would be obtained.26 26 The ISI search and retrieval facility Web of Science is an example where bibliographic coupling is applied for retrieval purposes. The output is a ranked list of documents bibliographically coupled with a test paper ( Po ), using Kessler’s original expression. In comparision with a ranked list, clusters of articles would in addition contain the information inherent in the links between the articles and also provide an over view of current research themes. 49 CHAPTER 3: RATIONALE AND RESEARCH DESIGN Based on the theoretical foundation presented in the previous Chapter 2, the rationale for the research design and the questions that it addresses are given in this chapter. 1. RESEARCH SETTINGS In order to reach a comprehensive understanding of the proposed method’s applicability as a science mapping tool, it was decided that it be tested under different environments. This is motivated by the fact that citation behavior, as reflected by lengths of reference lists and time-lag between publication and citation, differs between different fields (cf. Small & Sweeney, 1985), which affect the strength, number of bibliographic coupling linlcs and the density of the citation network. Though the precise choice of fields was of no immediate importance, a variation with regard to size of fields (publication output), referencing character and subject matter was strived for. It was also deemed important to test the proposed method for core document mapping. This is motivated by findings in previous researches where the important role of core documents in the science communication system has been established. This implies a large scale multidisciplinary research setting as the incidence of core documents generally should be low when delimited to a single field. From another viewpoint, it is also of interest to evaluate the proposed method in a research setting that is not restricted in terms of discipline borders (cf. Weinberg, 1974 and Vladutz & Cook, 1984). Such a research setting would give access to an immensely larger network, covering also cognitive relations transcending discipline borders. Based on the aforesaid reasons, it was decided that empirical tests be carried out in four different research settings. Each of the first three corresponds to a certain field of research. The three fields were: i. scientometrics; ii. organic chemistry; and iii. pure and applied mathematics. The fourth setting is on a multidisciplinary basis comprising an annual volume of the SCI. For convenience, henceforth, these four research settings are referred to as cases, Case 1 to Case 4. 2. RATIONALE AND RESEARCH QUESTIONS A basic assumption underlying the design of this study is that cluster analysis based on bibliographic coupling can not identify the cognitive cores of research specialties in the same direct fashion as cocitation cluster analysis, as there generally exists no clear indicator of document impact for currently published research articles. It is assumed, however, that clusters based on bibliographic coupling do reflect research specialties, though the extent to which the core of a specialty is identified can not be estab lished on basis of the aforesaid reason. Therefore, the term “current research themes” is regarded a more proper expression of what clusters in fact may reflect. This also brings 50 about that the design of this study is delimited to establish if the proposed method is capable of identifying such current research themes, rather than capable of elaborating the cognitive structures of specific specialties. By way of introduction two main purposes for science mapping were mentioned: (1) the study of the specialty structure of science and (2) scientific information provision. Without excluding (1) as a proper area of study with regard to cluster analysis based on bibliographic coupling, the design of this study aims at the evaluation of the proposed method in the context of scientific information provision. The evaluation of the proposed method has somewhat different aims with regard to the four cases. As for Cases 1 to 3, the objective is to test the proposed method’s applicability as a tool for the mapping of a particular field of research. In Case 4, the objective is to test the proposed method in a multidisciplinary research setting with a specific focus on the mapping of core documents. The rationales and research questions pertaining to Cases 1 to 3 and to Case 4 are presented in two separate sub-sections. 2.1 Cases 1 to 3 There are two major aspects in the evaluation of the proposed method with regard to Cases 1 to 3, namely, the relevance of cluster compositions and the agreement between intellectual-manual partitions of article populations performed by field experts and partitions generated by the complete link cluster method. The field experts’ partitions will be considered as an external point of reference in this study. Should both partitions generally agree, then something similar to an expert system27 would have been accomplished. However, little new information would be generated. On the other hand, should there be a general disagreement between two partitions, it would either indicate a failure to generate relevant clusters that are cognitively coherent, or, the adding of new information in terms of different but generally relevant clusters. The relevance of cluster compositions will be assessed by field experts where the number of documents that are not in line with the identified research theme of a cluster is established by way of inspection of articles in clusters. 27 A computer program that performs a task that would, otherwise, be performed by a human expert. The agreement between partitions generated by the complete link cluster method and the intellectual-manual method will be assessed through pair-wise comparisons of partitions. The following variables will be compared: i. the concentration of articles to clusters; ii. the internal coherence of clusters; and iii. the external isolation of clusters. 51 Concerning (i), the extent to which a partition leads to a dispersion of articles to many clusters or a concentration of articles to a few clusters, may reflect fragmentation as well as the amalgamation of research themes. With regard to (ii) and (iii), the internal coherence and the external isolation of a cluster reflect the extent to which a cluster is consistent and demarcated with regard to the definition of similarity applied for the merging of articles to clusters. Deviations with regard to (ii) and (iii) between partitions generated by the complete link cluster method and field experts respectively may reflect that semantic relations between documents (as perceived by field experts) are not mirrored by consensual referencing, or, that the field experts’ perceptions of cognitive resemblance between documents rely on intellectual classification schemes that supersedes both consensual referencing and semantic similarity. However, primarily the nature of the deviation between the complete link cluster method and the expert clustering is explored, not its causes. Deviations between partitions will also be assessed with regard to the composition of clusters. Here, a qualitative approach will be applied where the dispersion of articles over clusters will be studied by visual inspection. The research design covering Cases 1 to 3 aims to answer the following questions: QI. To what extent does the proposed method generate relevant clusters? Q2. What is the nature and extent to which results generated by the proposed method deviate from results generated by intellectual-manual partitions performed by field experts? Q3. What are the effects of the application of the proposed method with regard to applied thresholds and method of partition on document populations? Q4. What are the implications of the results in QI to Q3 with regard to the application of the proposed method as a tool for the mapping of science fields? 2.2 Case 4 There are three factors which motivated the research design of this case. They are: i. the incidence of core documents; ii. the properties of core documents; and iii. the properties of the complete link cluster method. With regard to (i), as was mentioned earlier, the incidence of core documents should generally be low for a single field. Glänzel and Czerwon (1996) found that less then one percent of all items (4,534 documents; in the 1992 volume 52 of SCI were core documents. These were dispersed over 42 sub-fields and assigned a total of 128 journal subject categories. This dispersion of core documents over a large number of fields and specialties underlines the necessity of a multidisciplinary research setting. With regard to points (ii) and (iii), considering the severe rule for merging of objects when the complete link cluster method is applied (see Sub-sections 1.4 1 and 1.4.2 in Chapter 2) and the role of core documents as central nodes in networks of bibliographically coupled articles with many and strong links to other articles, one could on theoretical grounds presume that core document clusters frequently would be parts of larger groups of related articles. In order to further elaborate the implications of this presumption, a strategy of mapping was outlined and will be applied. The strategy has its point of departure in a set of clusters generated by a first partition of the population of core documents. Here, only strong links will be used for the clustering of core documents. This partition forms a base line from which two lines of mapping will be pursued: i. In the first line of mapping, all significant (strong) links connecting core documents in clusters with any other core document will be mapped. This will result in a depiction of all significant artificially broken links between core documents in a cluster and core documents extrinsic to that cluster. The rationale for carrying out this line of mapping is that it will enable one to measure the extent of fragmentation of research themes the application of the proposed method may give rise to. ii. The second line of mapping involves the application of links between clusters only. They will be used to successively merge clusters on two subsequent levels of fusion, where the first generation of clusters are considered objects for a second clustering, and the second generation of clusters will give rise to a final cluster fusion. The rationale for carrying out this second line of mapping is that larger specialties with complex internal structures may be mapped when the information in links between clusters is applied. The impact of iterated clustering will be regarded with respect to the overall cluster structure, with a starting point at the base line. Changes of cluster composition on the three levels will be evaluated with regard to the following variables: i. the internal coherence of clusters; ii the external isolation of clusters; iii. the reduction of the number of clusters; iv. the increment of cluster sizes; 53 V. the number of isolated clusters; and vi. the number of singleton clusters. Concerning points (i) and (ii), the internal coherence and the external isolation of a cluster reflect the extent to which a cluster is consistent and demarcated with regard to the definition of similarity applied for the merging of articles to clusters. Points (iii) to (vi) would reflect effects that a' priori could be expected when applying iterated clustering. With regard to the multidisciplinary aspect of this research setting, a comprehensive expert evaluation of cluster relevance would be impracticable. Therefore, the assessment of cluster relevance (i.e. cluster subject coherence) will basically be grounded on statistical assessment of cluster properties. At the base line (first clustering), clusters are assumed to be subject consistent. This was deemed reasonable on the following grounds: i. Previous researches (e.g. Vladutz & Cook, 1984; Peters, Braam & van Raan, 1995) have shown that strong bibliographic coupling links between research articles generally imply subject relatedness. ii. Only strong links between core documents will be applied in the clustering. iii. The use of the complete link cluster method will exclusively generate completely interconnected clusters. iv. In consideration of (i) to (iii) above, subject coherent clusters should be expected. It is therefore, presumed that changes of cluster coherence will generally mirror changes of cluster subject coherence. Tikewise, changes of the external isolation are presumed to mirror the continuation or discontinuation of a specialty over levels of cluster fusion. In order to complement findings, four cases of iterated clusterings will be presented to field experts, who will be invited to evaluate and comment on the subject coherence and separation of clusters in terms of cluster relevance on different levels of cluster fusion. The selection of these cases is aimed at finding examples from the dominant scientific fields, namely physics, chemistry and bio-medical sciences. The design of Case 4 attempts to answer the following questions: QI. To what extent does the proposed method impose a fragmentation of specialties, when applied for core document mapping? Q2. What is the impact of iterated clustering on the overall cluster structure? Q3. Is there an optimal level of cluster fusion? 54 Q4. What are the implications of the results in QI to Q3 with regard to the application of the proposed method on core document data? 55 CHAPTER 4: METHODS AND DATA This chapter is divided into three sections. The first section is about the basic components of the proposed method which deals with measures of document similarity, the application of the complete link cluster method, the motives for the causal use of the between groups average cluster method and a minor experimental comparison of three agglomerative hierarchical methods. The next section describes the methods used in the evaluation of the proposed mapping method. The last section presents the process of data selection, properties of the document populations in the four research settings and a discussion of threshold settings and periods of observation. The latter is supplemented with a minor experiment. 1. THE BASIC COMPONENTS OF THE PROPOSED METHOD 1.1 Measurement of Proximity The first task when the objective is to partition sets of objects for mapping purposes is to find a method for deciding the proximity between objects. A proximity value is a number which indicates how similar or dissimilar two objects are. Proximity measures could be of two types as follows: i. Similarity measure - a function that maps the association between two objects so that the stronger the association, the higher the number; and ii. Dissimilarity measure - a function that maps the association between two objects so that the stronger the association, the' lower the number. In the context of cluster analysis, the notions of similarity and dissimilarity have a correspondence to distances. When applying a proximity measure that is a similarity measure, a high number corresponds to a small distance. A dissimilarity measure, on the other hand, yields the opposite result, i.e., a high number corresponds to a large distance. Generally, the similarity or dissimilarity between two objects can be measured in two essentially different ways as follows: i. Local approach - the direct similarity between two objects; or ii. Global approach - the way the objects relate to other objects in the population studied (Ahlgren, Jarneving & Rousseau, 2003). Where bibliometric studies are concerned, the local approach has frequently been applied in document cocitation cluster analysis, e.g. in the original works by ISI researchers (e.g. Small, 1973; Small & Griffith, 1974; Griffith, Small, Stonehill & Dey, 1974; Small & Griffith, 1983; Small & Sweeney, 1985) and also in research in bibliographic coupling (e.g. Sen & Gan, 1983; Vladutz & Cook, 1984). In other instances, the global approach has been the prevalent approach, as in author cocitation analysis, where the common method is to 56 compare an ordered pair of vectors of author cocitations and calculate Pearson’s r between these vectors (e.g. White & Griffith, 1981; McCain, 1990; White & McCain, 1998). Also, in many applications, objects are assumed to be represented as points in Euclidean space. The problem with the global approach is that sometimes the underlying data (e.g. values of bibliographic coupling strength or cocitation coupling strength) include associations of the type “co-absence”. “Co-absence” refers to pairs of objects which might be seen as similar in the sense that both objects lack an association to other objects in the population of study. This might in some instances correctly reflect the similarity between two such objects, and in other instances, it might not. Thus, one must ask if “co-absence” contain useful information about the similarity between two objects (Everitt, Landau & Leese, 2001, p. 36). In Leydesdorff (1987), the author points out that a large amount of “missing values” (zeros in a data matrix) could be potentially problematic when applying the Euclidean metrics to citation data: Since missing values do not add to the Euclidean distance between two cases, those cases with large amount of missing values end up with small distances among them, and when this is the clustering criterion (as for example in the single link clustering) clustering starts at this end. Hence, the possibility to cluster two objects on the ground of their difference from the rest of the set of objects, rather than on the ground of their similarity with each other is obvious. In this study, when performing some preliminary tests where Euclidean distances were used as input data to different cluster routines, this frequently brought about early fusions of documents on the basis of a small or zero distance in combination with a low coupling strength or the complete absence of couplings. This outcome was also found when Pearson’s r was applied as a proximity measure. In spite of this, the global approach seems appealing from the perspective that more information is underlying the values of similarity or dissimilarity. Due to this flaw of global measures, a local measure might be preferred. However, the original definition of bibliographic coupling strength (see Sub­ section 2.1 in Chapter 2) may not represent the optimal measure of document similarity. Also, the fact that two articles have a reference in common is no guarantee that both articles are referring to the same piece of information in the cited article (cf. Martyn, 1964). In addition, the significance of a reference is not known, and references may differ in terms of their impact on the citing article (see the comment on point (v) in Sub-section 3.3 in Chapter 2). In spite of this, it still seems reasonable to assume that the probability of the cognitive relationship between two documents should increase by the number of common references. Moreover, the significance of a bibliographic coupling unit associating two articles should be inversely related to the combined lengths of the reference lists of both documents (Vladutz & Cook, 1984). Therefore, a function that normalizes for the length of reference lists is needed. This calls for the use of the C.A. presented in Sub-section 2.1 in Chapter 2 (equation 2.3). The C.A. has been applied by several other researchers in the 57 past (e.g. Sharada & Sharma, 1993; Mubeen, 1995; Glän2:el & Czerwon, 1995 and 1966). For the sake of simplicity, this measure is defined here as: y ( r" ..a ’ (4.1) where NCSjj = the normalized coupling strength between article z and article j r,j = number of references common to both z and j rij = number of references in the reference list of article z rij = number of references in the reference list of article j The interval is [0, 1] and n¡ = rij = r¡¡ gives the maximum value. This function will be referred to as the Normalized Coupling Strength (NCS) henceforth in this study. 1.2 Application of the Complete Link Cluster Method Since all agglomerative hierarchical techniques reduce data to a single cluster containing all the objects, the search for an optimal number of clusters demands a decision of when to stop. Usually partitions an; achieved by cutting a dendrogram at a particular height, a “best cut” (Everitt, Landau & Leese, 2001, p. 76). This requires that clear shifts of fusion levels are discernable. However, in no case did a marked hierarchical structure show up. This phenomenon is illustrated in Figure 4-1. 58 Figure 4-1: The Distribution of Clusters at Different Fusion Coefficients 200 180 160 tn 140 ÿ 0 120 =o loo oL. 0.25, which resulted in a reduction to 267,034 links. In these links, 6,060 unique core documents were identified and constituted a final set for the analysis. 41 Hence, it was considered sufficient for the purpose of the study to only include items that could be categorized as genuine research articles. 2 The JCR is a multidisciplinary journal citation database launched by the ISI, providing means for the evaluation the impact of scholarly journals on research. It covers more than 7,500 of the world's most highly cited, peer-reviewed journals in approximately 200 disciplines. It provides two editions: the Science Edition and the Social Science Edition. The dispersion of articles over disciplines was assessed by computing the distribution of core documents over journal subject categories assigned by the ISI. It was found that most articles were published in journals assigned more than one subject category. Hence, the topic of each journal (or article) only approximates the combined subject categories assigned to it. These compound classification codes or strings were counted and a total of 379 unique strings were found. In this sense, a total of 379 unique classification codes were assigned to the set of core documents. When counting each unique subject category, a total of 129 were found, which approximates 76 percent of all subject categories in the Journal Citation Report, Science Edition.42 76 The more frequent subject categories are presented with the share of core documents in which they appear in Table 4-7. As can be seen, physics dominates followed by bio-sciences. Table 4-7: The Distribution of Core Documents over Journal Subject Categories in Case 4 Share of Core Documents Subject Categories 11% Physics, Applied 10% Physics, Multidisciplinary 10% Physics, Condensed Matter 9% Biochemistry & Molecular Biology 7% Physics, Particles & Fields 5% Materials Science, Multidisciplinary 5% Crystallography 4% Optics 3% Physics, Atomic, Molecular & Chemical 3% Cell Biology 3% Chemistry, Physical 3% Immunology 3% Engineering, Electrical & Electronic 3% Cardiac & Cardiovascular Systems 3% Oncology 3% Endocrinology & Metabolism 2% Surgery 2% Hematology 2% Multidisciplinary Sciences 2% Physics, Mathematical 2% Genetics & Heredity 2% Chemistry, Multidisciplinary 2% Neurosciences 2% Physics, Nuclear 2% Polymer Science Note: i. Subject categories assigned to less than two percent of the set of core documents are not shown. ii. In total, 129 subject categories were found. iii. Subject categories are in most cases overlapping. 77 CHAPTER 5: FINDINGS In this chapter, results from the empirical experiments are presented in accordance to case order, Case 1 to Case 4. For all cases, effects of the applied thresholds on the original populations as well as features of the same are presented in the previous Chapter 4 Section 3 in order to keep the issues of data selection, features of data, thresholds settings and features of the final document populations assembled. Hence, for each case findings are reported with a starting point in the final populations ready for clustering. 1. CASE 1 : SCIENTOMETRICS 1.1 Clusters Generated by the Complete Link Cluster Method A total of 185 articles from the original population of 232 articles were merged, resulting in 92 clusters of which 28 were singleton clusters. Approximately 34 percent of all articles were grouped in clusters containing at least three articles (see Table 5-1). These were selected for further analysis. The bibliographic descriptions of articles in these clusters are presented in Appendix 2. Table 5-1: The Size-Frequency Distribution of Clusters Generated by the Complete Link Cluster Method Size Frequency Size • Frequency Percentage 8 1 8 4 4 7 28 15 3 9 27 15 2 47 94 51 1 28 28 15 Sum 92 185 100 1.1.1 Coherence and Separation For the 17 selected clusters containing 63 articles, the AvgCS(C) (see equation 4.2) was measured. The median AvgCS(C) was 3.67 and the shape of the resulting distribution is shown in Figure 5-1. 78 Next, the separation between clusters was measured as the AvgCS(C, C) (see equation 4.3). Of all 2-combinations of clusters, 42 percent were coupled and no cluster was completely isolated. The median AvgCS(C, C) was 0.19 and the shape of the resulting distribution is shown in Figure 5-2. Figure 5-1: The Distribution of Coefficients of AvgCS(C) Fr eq ue nc y 10 6— 4- 8' 2— 0—|------------- ------------ -j------------- ------------- --------------------------- ------------- 2,00 4,00 6,00 8,00 10,00 12,00 14,00 16,00 AvgCS(C) Q © Figure 5-2: The Distribution of Coefficients of AvgCS(C, C) 30— 20— 81 3,0— 2,5— 2,0- 1,5— 1,0 0,5— 0,0- 0,20 0,80 o c 0> 3 CT 3 generated by the complete link cluster method will be referred to as COMP and the set of clusters generated by the field experts partitions as EXP. 1.3.1 The Coherence of Clusters With regard to AvgCS(C), both distributions are positively skewed, and the median value is higher for COMP. The median of D for EXP is 0.61, which should be compared with the default value of 1.0 for COMP. Conclusively, more coherent clusters were generated by the complete link cluster method. 1.3.2 The Separation Between Clusters With regard to AvgCS(C, C), both distributions are strongly positively skewed as most values are clustered at the lower intervals of the scale but the median for COMP is somewhat higher. The share of 2-combinations of clusters that are coupled is considerably lower for COMP. Conclusively, the general level of connectedness between clusters is lower for COMP, but links connecting clusters in COMP are slightly stronger. 1.3.3 The Concentration of Articles to Clusters The two partitions differ much as the number of clusters in COMP was higher, 17 vs. 10 clusters in EXP. For COMP the value of Pratt’s measure of concentration (see equation 4.4) is 0.13 and for EXP 0.39. Conclusively, articles are less concentrated to clusters in COMP. 1.3.4 The Qualitative Assessment of Cluster Compositions A couple of clusters are near identical, and eight clusters in COMP constitute subsets of clusters in EXP. Two EXP clusters are completely split up by pairs of COMP clusters. The general pattern is that clusters in EXP are split up by two or three clusters in COMP. Conclusively, when compared, clusters in EXP are fragmented by clusters in COMP and two much deviating partitions are seen. For a detailed comparison of COMP and EXP with regard to the cluster compositions, see Appendix 3. 83 1.4 The Field Expert’s Evaluation The field expert performed a visual inspection of the set of clusters generated by the complete link cluster method, examining all articles in order to detect inconsistencies as to subject content in clusters. Any article deviating from the major subject theme of a cluster was regarded as misplaced and marked. In total, six articles (10 percent), all elements in two clusters, were marked as misplaced. Comparing EXP with COMP, the field expert’s view was that some deviations between partitions could be “renegotiated”, foremost in terms of splitting up expert clusters. 1.5 Summary of Findings in Case 1 Below is a summary of the findings for this case. i. More coherent clusters were generated by the complete link cluster method. ii. Clusters generated by the complete link cluster method were less interconnected in terms of link density, though links between clusters were generally slightly stronger. iii. The concentration of articles to clusters was lower for clusters generated by the complete link cluster method, which also was reflected by more and smaller clusters. iv. Clusters generated by the complete link cluster method split up clusters generated by the field expert and presented a more fragmented picture of the analyzed research field. v. Clusters generated by the complete link cluster method were generally relevant. The basic features of the two deviating structures of clusters are depicted by MDS as graphs in Figures 5-6 and 5-7, where the distances between clusters are based on the AvgCS(C, C). Here, circle sizes are proportional to cluster sizes and the width of connecting links to the strength of AvgCS(C, C). These are not directly comparable between maps as there may be some scale difference. The sizes of a largest and a smallest cluster are given in the notes below the figures for guidance. 84 Figure 5-6: Graph of the 17 Clusters Generated by the Complete Link Cluster Method Visualized by MDS 4 Note: Kruskal’s stress is 0.12. N for cluster 51 is 8 and N for cluster 48 is 3. 85 Note: Kruskal’s stress is 0.14. N for cluster 5 is 10 and N for cluster 10 is 1. Figure 5-7: Graph of the 9 Clusters Generated by the Field Expert Visualized by MDS 86 2. CASE 2: ORGANIC CHEMISTRY 2.1 Clusters Generated by the Complete Link Cluster Method A total of 268 articles from the original population of 14,389 articles were merged, resulting in 95 clusters of which 17 were singleton clusters. Approximately 68 percent of all articles were grouped in clusters containing at least three articles (see Table 5-4). These were selected for further analysis. The bibliographic descriptions of articles in these clusters are presented in Appendix 4. Table 5-4: The Size-frequency Distribution of Clusters Generated by the Complete Link Cluster Method Size 8 Frequency 1 Size • frequency 8 Percentage 3“ 7 2 14 5 6 3 18 7 5 11 55 21 4 7 28 10 3 20 60 22 2 34 68 25 1 17 17 6 Sum 95 268 100 2.1.1 Coherence and Separation For 44 selected clusters containing 183 articles, the AvgCS(C) (see equation 4.2) was measured. The median AvgCS(C) was 13.94 and the shape of the resulting distribution is shown in Figure 5-8. 87 Figure 5-8: The Distribution of Coefficients of AvgCS(C) Fr eq ue nc y 4— 6— 2 8- 10,00 25,0015,00 20,00 AvgCS(C) o-f— 5,00 u □ Figure 5-9: The Distribution of Coefficients of AvgCS(C, C) Note: Isolated clusters are not included. 10,00 AvgCS(C, C) 88 2.2 Core Documents - a Microanalysis As the size and character of the field allows for the identification of core documents, a microanalysis of the application of the proposed method’s applicability for core document mapping was pursued. In the network of shared references, a total of 49 source articles fulfilled the requirements for core documents as defined (see Sub-section 2.1 in Chapter 2). In this set, all but two articles were bibliographically coupled with another core document. Computing bibliographic couplings between articles in this set, the exclusion of pairs sharing less than two references resulted in a total of 186 coupled pairs, 2,333 bibliographic couplings and 46 articles. The range of NCS was 0.88 (i.e. 0.024 - 0.90) and the median coefficient was 0.30. A matrix, containing "46" .2, = 1,035 elements based on this set was computed, which could be considered a rather dense subgraph where the density D (see equation 2.2 ) was 0.17.43 This matrix was then used as input for complete link clustering and MDS. It showed up as seven clearly demarcated clusters on the MDS map in full agreement with clusters generated by the complete link cluster method (see Figure 5-14 in Sub-section 2.5 in this Section). 43 This could be compared with the share of 2 percent of 2-combinations of articles that were coupled in the original matrix. 2.3 Clusters Generated by the Field Expert Applying the card sorting technique, the field expert performed an intellectual partition of the articles contained in the 44 clusters with the minimum size of three articles, generated by the complete link cluster method. 2.3.1 The Partition The field expert in this case used a principle of division based on four major categories: i. properties; ii. synthesis; iii. understanding of mechanism; and iv. method. These four categories were in turn subdivided. The rectangular table used for the card sorting was applied in a way that a spatial representation was accomplished (see Figure 5-10). 89 Figure 5-10: The Configuration of Piles on the Card Sorting Table 10 12 14 13 15 17 The configuration of 17 piles of cards of bibliographic representations of articles were placed in a way that piles (clusters) with a similar or connecting research foci were located in each others vicinity and each pile was assigned a label indicating the perceived research focus (see Table 5-5). Table 5-5: The Field Expert’s Labels Cluster Labels 1 Synthesis product 2 Synthesis reaction 3 Catalysis 4 Stereo selective synthesis-product 5 Stereo selective synthesis-reactions 6 Stereo selective synthesis-catalysis 7 Total synthesis 8 Total synthesis & medicinal chemistry 9 Stereo selective synthesis-racemization 10 Synthesis evaluation 11 Peptide synthesis 12 DNA templated organic synthesis 13 Stereo selective reaction mechanism 14 Reaction mechanism 15 DNA properties 16 Peptide structure 17 Nano All articles were contained in clusters with a size of at lea st three articles, with the exception of three singleton clusters. The distribution of articles over clusters is shown in Table 5-6. 90 Table 5-6: The Size-frequency Distribution of Clusters Generated by the Field Expert’s Clustering Size Frequency Size • Frequency Percentage 43 1 43 23 29 1 29 16 25 1 25 14 14 2 28 15 13 1 13 7 12 1 12 7 10 1 10 5 5 1 5 3 3 5 15 8 1 3 3 2 Sum 17 183 100 2.3.2 Coherence and Separation The coherence of the field expert’s clusters in terms of bibliographic coupling units between articles in clusters was measured as the AvgCS(C) and D (see equation 2.2). The median AvgCS(C) was 3.91 and the median D 0.30. The shapes of the resulting distributions are shown in Figures 5-11 and 5-12. Figure 5-11: The Distribution of Coefficients of AvgCS(C) Fr eq ue nc y 4- 5' 3— 2- 1- 15,00 0----- 0,00 5,00 10,00 AvgCS(C) CD 0) CD Note: Singleton clusters are not included. 91 Figure 5-12: The Distribution of Coefficients of D Fr eq ue nc y o> 4- 5' 2— 3- 1- 0,20 0,80 1,00 0----- 0,00 0,40 0,60 D o o> Note: Singleton clusters are not included. The separation between clusters was measured as the AvgCS(C, C). Of all 2- combi nations of clusters, 35 percent were coupled and four clusters were found completely isolated. The median AvgCS(C, C’) was 0.40 and the shape of the resulting distribution is shown in Figure 5-13. Figure 5-13: The Distribution of Coefficients of AvgCS(C, C) Fr eq ue nc y 25— 30— 20— 1 1 5— 0_|---- ----------1---- 1---- 1---- J--------- 1---- I---- (--------- 0,00 1,00 2,00 3,00 4,00 5,00 6,00 AvgCSfC, C) oc 3 generated by the complete link cluster method will be referred to as COMP and the set of clusters generated by the field experts partitions as EXP. 2.4.1 The Coherence of Clusters With regard to AvgCS(C), the distribution of COMP is more normally distributed, whereas the distribution of EXP is slightly positively skewed. The two distributions are deviating, as the median of COMP is more than three times the median of EXP. The median of D for EXP is 0.30 which should be compared with the default value of 1.0 for COMP. Conclusively, clusters in COMP are generally much more coherent. 2.4.2 The Separation between Clusters Both COMP and EXP have isolated clusters, five and four respectively. With regard to AvgCS(C, C), both distributions are strongly positively skewed with extremes. For COMP, the extremes are more frequent and clustered at a higher level of the scale, indicating relatively strong associations between several clusters. Also, the median AvgCS(C, C) for COMP is lower. The share of 2- combinations of clusters that are coupled is considerably lower for COMP too. Conclusively, the general level of connectedness between clusters in COMP is lower, however, a considerable share of relatively strong links (extremes) between more than 20 clusters was found in the case of COMP (all extreme links can are reflected as clusters of clusters in the map in Figure 5-15 shown under Sub-section 2.5 in this chapter). 2.4.3 The Concentration of Articles to Clusters The two partitions differ much as the number of clusters in Comp was higher, 44 vs. 17 clusters in EXP. This involves a difference between the two partitions with regard to the concentration of articles to clusters. The value of Pratt’s measure of concentration (see equation 4.4) is 0.17 for COMP and 0.57 for EXP. Conclusively, articles were more concentrated to clusters in the case of EXP. 93 2.4.4 The Qualitative Assessment of Cluster Compositions The most striking deviation is the difference in cluster sizes which is pronounced by three large expert clusters with a size between 43 and 25. Over these three clusters, approximately half of all articles are dispersed and the number of intersections with COMP is 53. Several (14) COMP clusters are completely contained within an EXP cluster, and approximately 30 percent of all articles belong to this category. Therefore, several of COMP’s clusters constitute fractions of EXP’s clusters. It can be concluded that two very deviating classification systems have been operating. For a detailed comparison of COMP and EXP with regard to the cluster compositions, see Appendix 5. 2.5 The Field Expert’s Evaluation When evaluating the partition generated by the complete link cluster method, the field expert concluded that “there exists several possible common denominators” for a single cluster solution, and these were often hard to decide. In fact, the proposed method supplied the expert with a new set of principles of division which were not easy to anticipate. Some of the classes brought about by the complete link clustering could clearly be of interest in relation to several research questions at issue. A number of clusters were based on the association between articles with a common focus on specific methods whereas others seemed to be based on the association between articles with a common focus on chemical compounds or classes of compounds. Regarding the evaluation of clusters, all 44 clusters were inspected article by article in order to detect inconsistencies as to subject content in clusters where any article deviating from the major subject theme of a cluster was regarded as misplaced and marked. In total, four articles in four different clusters were regarded as deviating from the cluster’s research focus. Concerning the map of core documents, the expert performed an evaluation of both the clusters and the configuration of the map (see Figure 5-14). One article (13439) was considered unclassifiable, as the expert was not familiar with the subject presented by this article. Except for this article, all other articles were considered correctly located. In the configuration in the MDS map, two dimensions could be discerned: i. Top-bottom “Non-stereo selective reactions” to “stereo selective reactions". ii. Left-right “Metal-catalysis” to “other chemistry”. The configuration of clusters in the MDS map was considered intelligible, though cluster 1 should be more distant to the other clusters as its research 94 theme (DNA) was divergent in relation to the research themes of all the other clusters in the map. The bibliographic descriptions of evaluated core documents are presented in Appendix 6. Figure 5-14: Map of the Associations between 46 Core Documents Generated by MDS and the Complete Link Cluster Method Cluster 1 Cluster 2 '2i :014 Cluster 3 Cluster 4 9712 Cluster 5 lusa Cluster 6 Cluster 7 3394 3676 / 6342 3739 \ 1703 3493 2746 596? 3605 12554 4502 ' 347 13439 779 262 11273 S 6095 7743 \ 2516 4031 oGS9g30Q 5305 10681 i 8335 / 1142 72579 1«.f"í4 1089 Note: Kruskal’s stress is 0.03. 95 2.6 Summary of Findings in Case 2 Below is a summary of the findings for this case. i. More coherent clusters were generated by the complete link cluster method. ii. Clusters generated by the complete link cluster method were less interconnected in terms of link density, and generally, links were also weaker. However, several clusters generated by the complete link cluster method were associated by strong links (extremes). iii. The concentration of articles to clusters was weaker for clusters generated by the complete link cluster method, which also was reflected by more and smaller clusters. iv. Clusters generated by the complete link cluster method presented a more fragmented picture of the analyzed research field and there is little agreement between the two partitions. v. Clusters generated by the complete link cluster method were mostly completely relevant. vi. The clustering and mapping of core documents id entified and depicted highly relevant clusters and the configuration of clusters when mapped was basically relevant. The basic features of the two deviating structures of clusters are depicted by MDS as graphs in Figures 5-15 and 5-16. where the distances between clusters are based on the AvgCS(C, C). Here, circle sizes are proportional to cluster sizes and the width of connecting links to the strength of AvgCS(C, C). These are not directly comparable between maps as there may be some scale difference. The sizes of a largest and a smallest cluster are given in the notes below the figures for guidance. 96 Figure 5-15: Graphs of the 39 Clusters Generated by the Complete Link Cluster Method Visualized by MDS ¿41 42. 4 j33 32| KX 9J 3/ Note: Clusters 4, 6, 23, 40 and 43 are not mapped as they are isolated. N for cluster 18 is 8 and N for cluster 35 is 3. Kruskal’s stress is 0.04. 97 Figure 5-16: Graph of the 13 Clusters Generated by the Field Expert Visualized by MDS 2 Note: Clusters 11, 12, 15 and 17 are not mapped as they are isolated. N for cluster 2 is 43 and N for cluster 10 is 1. Kruskal’s stress is 0.13. 98 3. CASE 3: PURE & APPLIED MATHEMATICS 3.1 Clusters Generated by the Complete Link Cluster Method A total of 579 articles from the original population of 879 articles were merged, resulting in 420 clusters of which 311 were singleton clusters. Approximately 22 percent of all articles were grouped in clusters containing at least three articles (see Table 5-7). These were selected for further analysis. The bibliographic descriptions of articles in these clusters are presented in Appendix 7. Table 5-7: The Size-Frequency Distribution of Clusters Generated by the Complete Link Cluster Method Size Frequency Size • frequency Percentage 6 1 6 1 5 1 5 1 4 5 20 3 3 33 99 17 2 69 138 24 1 311 311 54 Sum 420 579 100 3.1.1 Coherence and Separation For 40 selected clusters containing 130 articles, the AvgCS(C) (see equation 4.2) was calculated. The median was 4.17 and the shape of the resulting distribution is summarized in Figure 5-17. 99 AvgCS(C) Figure 5-17: The Distribution of Coefficients of AvgCS(C) Fr eq ue nc y 12— 6- 9— 3- 0' 5,00 10,00 15,00 20,00 oc 3 generated by the complete link cluster method will be referred to as COMP and the set of clusters generated by the field experts partitions as EXP. 3.3.1 The Coherence of Clusters With regard to AvgCS(C), both distributions are positively skewed with some extremes. For EXP, the extremes refer to a couple of clusters with the size of two, which have similar subject foci and very similar reference lists. The median value of AvgCS(C) was considerably higher for COMP. With regard to D. the median for EXP was 0.33. which should be compared with the default value of 1.0 for COMP. Conclusively, more coherent clusters were generated by the complete link cluster method. 3.3.2 The Separation between Clusters With regard to AvgCS(C, C), both distributions are positively skewed with a few extremes each, and the median is lower for COMP. In addition, five clusters in COMP were isolated. The share of 2-combinations of clusters that are coupled is lower for COMP, though for both distributions holds that the general level of interconnectedness between clusters is low. Conclusively, the general level of interconnectedness between clusters is lower for COMP, and links connecting clusters in COMP are generally weaker too. 3.3.3 The Concentration of Articles to Clusters The two partitions differ with regard to the dispersion of articles over clusters. The articles contained in 40 clusters in COMP, are dispersed over 26 clusters in EXP and an additional ten articles are singleton clusters. The value of Pratt’s measure of concentration (see equation 4.4) is 0.07 for COMP and 0.42 for EXP. Conclusively, articles are less concentrated to clusters in COMP. 3.3.4 The Qualitative Assessment of Cluster Compositions Three clusters, all with the size of two, are identical, and eight clusters in COMP constitute subsets of clusters in EXP. Two EXP clusters are completely split up by pairs of COMP clusters. In all, clusters in COMP provide a more 105 fragmented depiction of the field and two much deviating partitions are seen. For a detailed comparison of COMP and EXP with regard to the cluster compositions, see Appendix 8. 3.4 The Field Expert’s Evaluation A visual inspection of the clusters generated by the complete link cluster method was performed by the field expert and 40 clusters were inspected article by article in order to detect inconsistencies as to subject content in clusters and any article deviating from the major subject theme of a cluster was regarded as misplaced and marked. A total of three articles were regarded as unclassifiable in the sense that titles and additional bibliographic information added insufficient information regarding topics. In total, 17 articles (13 percent) were marked as misplaced. 3.5 Summary of Findings in Case 3 Below is a summary of the findings for this case. i. More coherent clusters were generated by the complete link cluster method; ii. Clusters generated by the complete link cluster method were less interconnected in terms of link density, and links between clusters were generally weaker. iii. The concentration of articles to clusters was lower for clusters generated by the complete link cluster method. The dispersion of articles differed considerably between the two partitions, foremost with regard to the evenness of the distribution of cluster sizes where clusters generated by the complete link cluster method showed a lower variation. iv. The cluster structure generated by the complete link cluster method depicted the analyzed research field as more fragmented. v. Clusters generated by the complete link cluster method were generally relevant. The basic features of the two deviating structures of clusters are depicted by MDS as graphs in Figures 5-22 and 5-23, where the distances between clusters are based on the AvgCS(C, C). Here, circle sizes are proportional to cluster sizes and the width of connecting links to the strength of AvgCS(C, C). These are not directly comparable between maps as there may be some scale difference. The sizes of a largest and a smallest cluster are given in the notes below the figures for guidance. 106 Figure 5-22: Graphs of the 35 Clusters Generated by the Complete Link Cluster Method Visualized by MDS >38' 2S 2\ \2\ Note: Clusters 7, 21, 29, 35 and 37 are not mapped as they are isolated. N for cluster 18 is 6 and N for cluster 34 is 3. Kruskal’s stress is 0.03 107 Figure 5-23: Graph of the 36 Clusters Generated by the Field Expert Visualized by MDS 12. 2& 2; 2; 26 •9 Note: Kruskal’s stress is 0.07. N for cluster 24 is 16 and N for cluster 8 is 1. 108 4. CASE 4: CORE DOCUMENTS The findings of this case is presented in a different manner from the previous four cases in view of its complexity and different research questions. For simplicity purpose, the following notations will be used when referring to the different levels at which clusters were generated: i. Cl denotes the level at which clusters were generated by the first clustering. Applied measure of document similarity was NCS and the cluster method was the complete link cluster method. From here and onwards, clusters generated at the Cl level are referred to as “Cl” with an added cluster identification number (e.g. CI/32) when a particular cluster is referred to. When the complete identity of a Cl cluster is addressed, all levels are noted, e.g. C3/3/C2/87/C1/1931. ii. C2 denotes the level at which clusters were generated by the second clustering. Applied measure of cluster similarity was AvgCS(C, C) and the cluster method was again the complete link cluster method. From here and onwards, clusters generated at the C2 level are referred to as “C2” with an added cluster identification number when a particular cluster is referred to. iii. C3 denotes the third and last level of clustering. Applied measure of cluster similarity was AvgCS(C, C) and the method of partition was the between groups average cluster method. From here and onwards, clusters generated at the C3 level are referred to as “C3” with an added cluster identification number when a particular cluster is referred to. The findings are presented in the following order: 1. Results from the first clustering (C 1 -level) 2. Results from the iterated clustering (C2-level) 3. Results from the reiterated clustering (C3-level) 4. Field experts’ evaluations of four cases of reiterated clustering 5. Results from the expansion of Cl-clusters 6. Summary of findings Though the presentation of findings should aim at an order that is in line with the sequence of pursued experiments, for practical reasons, results from the clustering on the different levels of fusion are presented in one sequence in order to facilitate the comprehension of changes. However, point 5 preceded points 2 and 5 empirically. The fusions of clusters at the different levels are illustrated by examples. With the starting point in a Cl-cluster, the stepwise merging of clusters to a final 109 C3-cluster is described. The selection of the C3-cluster was based on its approachability on a layman level in terms of comprehensible core document titles.44 With regard to point 5, an example of an expanded Cl-cluster is also given. 44 The author of this thesis has a background in health sciences, and associations between articles and clusters could be interpreted, at least in a superficial way, as the general topic of the example cluster is about infectious diseases. 4.1 The First Fusion Level - Cl Clusters 4.1.1 Clusters and Cluster Sizes In order to assess the association through bibliographic coupling between core documents, the NCS between all 6,060 core documents of the original population was computed (see Sub-section 3.2.5 in Chapter 4). Links with a NCS lower than 0.25 were filtered out and 5,771 articles were clustered by the complete link cluster method. A total of 1,761 clusters were generated of which 228 were singleton clusters. In all, 5,543 core documents were merged to 1,533 clusters varying in size between 2 and 22. 1000 clusters had a size > 3 and contained in total 4,477 core documents (see Table 5-10). These 1000 clusters were selected for further analysis and fusion. Table 5-10: The Size-Frequency Distribution of Cl-Clusters Size Frequency Size ■ Frequency Percentage 22 2 44 0.76 20 1 20 0.35 15 1 15 0.26 14 1 14 0.24 13 1 13 0.23 12 3 36 0.62 11 11 121 2.10 10 10 100 1.73 9 14 126 2.18 8 22 176 3.05 7 46 322 5.58 6 88 528 9.15 5 139 695 12.04 4 284 1136 19.68 3 377 1131 19.60 2 533 1066 18.47 1 228 228 3.95 Sum 1761 5771 100.0 The shape of the distribution of core documents over C l-clusters where the size is > 3 is summarized in Figure 5-24. 110 4.1.2 Figure 5-24: The Distribution of Core Documents over 1000 Cl-Clusters Fr eq ue nc y 400— 200- 300— 100- 0_|.......... .. I 7 ».. ..1.... 1.. I .y .1.. 1-------------- !---------------------r------------------- 0,00 5,00 10,00 15,00 20,00 25,00 Number of core documents in clusters o a>□ 0) As can be seen from Figure 5-24, the distribution of core documents over clusters is positively skewed. The median cluster size is 4. Approximately 78 percent of all core documents are contained within these clusters and 22 percent in clusters of lesser size. Coherence and Separation For the 1000 Cl-clusters containing at least three articles, the AvgCS(C) (see equation 4.2) was calculated. The shape of the distribution is summarized in Figure 5-25. Ill Figure 5-25: The Distribution of Coefficients of AvgCS(C) over 1000 Cl Clusters Fr eq ue nc y 120- 100- 40— 60— 20— 80— 10,00 40,0020,00 30,00 AvgCS(C) o-j-J- 0,00 u 0) 0) As can be seen from Figure 5-25, the distribution is quite symmetrical and the mean AvgCS(C) is 10.59. With regard to the aspect of isolation, the AvgCS(C, C) (see equation 4.3) between 1000 core document clusters was calculated. Of these, 49 core document clusters were isolated. The shape of the distribution is summarized in Figure 5-26. 112 Note: 49 isolated clusters are excluded. Figure 5-26: The Distribution of Coefficients of AvgCS(C, C) over Links between 951 Cl-Clusters Fr eq ue nc y 2 500— 2 000- 3 000- 1 500- 1 000- 500— 15,00 o-p-J 0,00 5,00 10,00 AvgCS(C, C) oc Q345 (where the AvgCS(C, C) > 3.11), 1,628 links connecting 706 Cl-clusters are found. This shows that a large share of clusters on the Cl level is connected by relatively strong links. 45 Q3 denotes the third quartile, i.e., the score that divides the bottom three quarters of the distribution from the top quarter. 46 SARS is the acronym for Severe Acute Respiratory Syndrome. 4.1.3 Example of Cluster Fusion on the Cl-Level The fusion of core documents on the Cl level is illustrated by cluster CI/1391. In this cluster, the focus is on SARS46 and all eleven constituent papers consistently treat this subject. The core documents in this cluster were published in ten different journals and assigned twelve different journal subject categories. The constituent articles are presented with article number, article title journal title and journal subject category as follows: i. 27178/ Chest-X-Ray Imaging of Patients with SARs / Chinese Medical Journal I Medicine, General & Internal ii. 28525/ Reovirus, Isolated from SARs Patients/ Chinese Science Bulletin / Multidisciplinary Sciences 113 iii. 110241/ Infection-Control for SARs in a Tertiary Neonatal Center/ Archives of Disease in Childhood/ Pediatrics iv. 219617/ Description and Clinical Treatment of an Early Outbreak of Severe-Acute-Respiratory-Syndrome (SARs) in Guangzhou, Pr- China/ Journal of Medical Microbiology/ Microbiology v. 275806/ A Clinicopathological Study of 3 Cases of Severe Acute Respiratory Syndrome (SARs)/ Pathology/ Pathology vi. 333109/ Severe Acute Respiratory Syndrome (SARs) - The Questions Raised by the Management of a Patient in Besancon and Strasbourg/ Presse Medicale/ Medicine, General & Internal vii. 383006/ Evaluation of WHO Criteria for Identifying Patients with Severe Acute Respiratory Syndrome Out-of-Hospital - Prospective Observational Study/ British Medical Journal/ Medicine, General & Internal viii. 400512/ Severe Acute Respiratory Syndrome-Associated Coronavirus Infection/ Emerging Infectious Diseases/ Immiunology; Infectious diseases ix. 400574/ Microbiologie Characteristics, Serologic Responses, and Clinical-Manifestations in Severe Acute Respiratory Syndrome, Taiwan/ Emerging Infectious Diseases/ Immunology; Infectious diseases X. 490401/ Safe Tracheostomy for Patients with Severe Acute Respiratory Syndrome/ Laryngoscope/ Medicine, Research & Experimental; Otorhinolaryngology xi. 527101/ Severe Acute Respiratory Syndrome in Hemodialysis- Patients-A Report of 2 Cases/ Nephrology Dialysis Transplantation/ Transplantation; Urology & Nephrology It can be seen from the above list of constituent articles that the problem of SARS is treated with a point of departure in several problem areas like diagnosis (both in vitro as well as in vivo), pathology, clinical treatment and specific clinical problems associated with this syndrome. It would also be interesting to know which cited works connect the core documents in cluster 1391 (see Table 5-11). As the size of this cluster is 11, cited works with a frequency of 11 would be common to all articles in this cluster. 114 Table 5-11: The Frequency of Works Cited by Articles in Cluster 1391 Frequency Cited Work 11 Drosten C. 2003, V348, P1967, New Engl J Med 11 Lee N, 2003, V348, P1986, New Engl J Med 9 Peiris JSM, 2003, V361, P1319, Lancet 9 Tsang KW, 2003, V348, P1977, New Engl J Med 8 Ksiazek TG, 2003, V348, P1953, New Engl J Med 7 Poutanen SM, 2003, V348, P1995, New Engl J Med 2 Ho W, 2003, V361, P1313, Lancet 2 Hon KLE, 2003, V361, P1701, Lancet 2 Li TST, 2003, V361, P1386, Lancet 2 Peiris JSM, 2003, V361, P1767, Lancet 2 WHO, 0000, CAS DEF Surv SEV AC 2 WHO, 0000, Cum Numb Rep Prob CA Note: Only cited works with a frequency > I are shown in the table. Three of the cited works with the highest frequencies are all published in the same journal and in the same issue (New England Journal of Medicine, 2003, V348, N20). In addition, they are all about the outbreak of SARS in Hong- Kong in the year 2003: i. Drosten C et al/ Identification of a Novel Coronavirus in Patients with Severe Acute Respiratory Syndrome ii. Lee N et al/ A Major Outbreak of Severe Acute Respiratory Syndrome in Hong-Kong Hi. Tsang KW et al/ A Cluster of Cases of Severe Acute Respiratory Syndrome in Hong-Kong Studying the compiled reference list of this cluster, the novelty value of the identified research theme on SARS and the identification of a research front issue is clear. 115 4.2 The Second Fusion Level - C2 Clusters 4.2.1 Clusters and Cluster Sizes On the basis of the AvgCS(C, C) between Cl-clusters containing at least three articles, 6.537 links connecting 951 Cl-clusters were applied for an iterated clustering. No threshold of AvgCS(C, C) was applied. The clustering of the 951 Cl-clusters resulted in 153 singleton clusters and 212 clusters varying in size between 5 and 97 articles (see Table 5-12). In total, 3,524 core documents were contained in the set of 212 C2-clusters. These were selected for further analysis and fusion. Table 5-12: The Size-Frequency Distribution of C2-Clusters Size Interval Frequency 90-99 1 80-89 1 70-79 0 60-69 2 50-59 5 40-49 5 30-39 7 20-29 29 10-19 98 0-9 64 Sum 212 Note: Singleton clusters are excluded and the class interval is 10. The shape of the distribution is summarized in Figure 5-27. 116 Figure 5-27: The Distribution of Core Documents over C2-Clusters Fr eq ue nc y 4» 40— 20— 60— 30— 70— 50- 10- 0' 60,0020,00 40,00 80,00 100,00 oc50 can be seen. These are foremost large physics clusters, but two are from the bio-medical sciences.47 47The following fields were presented by macro cluster in accordance to their sizes: Particle physics, Condensed Matter, Crystallography, Applied Physics and Endocrinology & Oncology. 4,2.2 Coherence and Separation For the 212 C2-clusters, the AvgCS(C) was calculated. The shape of the distribution is summarized in Figure 5-28. 117 AvgCS(C) Figure 5-28: The Distribution of Coefficients of AvgCS(C) Over 212 C-2 Clusters Fr eq ue nc y 25— 20— 30— 5- 0' 15,00 20,005,00 10,00 u 0)□ Q3 (where the AvgCS(C, C’) > 0.0625), 179 links connecting 145 C2-clusters were found. In all, it means that there is a drastic reduction of strong links and number of connected clusters on this second level of cluster fusion. 4.2.3 Example of Cluster Fusion on the C2-Level As an example, cluster CI/1391 (see Sub-section 4.1.3 in this section), is fused with four other Cl-clusters on the C2-level to C2/170. The fusion is illustrated in Table 5-13. 119 Table 5-13: The Fusion of Five Cl-Clusters to C2/170 AvgCS(C, C) No. Shared References C1- Cluster Cluster Size Ci- Cluster Cluster Size 5.42 65 1501 4 1513 3 5.14 185 1390 9 1501 4 5.00 135 1390 9 1513 3 4.27 188 1391 11 1501 4 4.03 133 1391 11 1513 3 4.00 392 1390 9 1391 11 3.60 72 1389 5 1501 4 3.56 160 1389 5 1390 9 3.47 52 1389 5 1513 3 3.11 171 1389 5 1391 11 As can be seen from Table 5-13, all clusters are associated with one another with a strength clearly above the median AvgCS(C, C) on the Cl-level, forming a complete graph on the C2-level. The mean AvgCS(C) of this graph is 4.25 and D is 1.0. Hence, the average connectedness is lower than the mean value for the whole set of C2-clusters though all constituent core documents are bibliographically coupled with one another. Though the average strength of links in the resulting C2 cluster is below the mean for the population of C2- clusters, a clear subject relatedness between constituent core documents and Cl-clusters is seen when tabulating and sorting core document titles in accordance to cluster affiliation (see Table 5-14). Table 5-14: Titles of Articles in Five Cl-Clusters Merged to C2/170 1389 Clinical Analysis of 45 Patients with Severe Acute Respiratory Syndrome 1389 I'he Role of Radiological Imaging in Diagnosis and Treatment of Severe Acute Respiratory Syndrome 1389 Initial Otolaryngological Manifestations of Severe Acute Respiratory Syndrome in Taiwan 1389 A Young Infant with Severe Acute Respiratory Syndrome 1389 Clinical Presentation and Outcome of Severe Acute Respiratory Syndrome in Dialysis Patients 1390 Zcurve-Cov - A New System to Recognize Protein-Coding Cienes in Coronavirus Genomes, and Its Applications in Analyzing SARs-Cov Genomes 1390 Prediction of Proteinase Cleavage Sites in Polyproteins of Coronaviruses and Its Applications in Analyzing SARs-Cov Genomes 1390 Maintaining Dental Education and Specialist Dental-Care During an Outbreak of a New Coronavirus Infection - Part 1 - A Deadly Viral Epidemic Begins 1390 Role of China in the Quest to Define and Control Severe- Acute-Respiratory- Syndrome 1390 A Hospital Outbreak of Severe Acute Respiratory Syndrome in Guangzhou. China 1390 An Outbreak of Severe Acute Respiratory Syndrome Among Hospital Workers in a Community-Hospital in Hong-Kong 1390 Epidemiology and Cause of Severe Acute Respiratory Syndrome (SARs) in Guangdong, Peoples-Republic-of-China, in February, 2003 1390 T ransmission Dynamics of the Etiologic Agent of SARs in Hong- Kong - Impact of Public-Health Interventions 1390 Children Hospitalized with Severe Acute Respiratory Syndrome- Related Illness in Toronto 120 Note: The first column holds the numbers of constituent Cl-clusters in C2/170. Table 5-14 continued... 1391 Severe Acute Respiratory Syndrome-Associated Coronavirus Infection 1391 Microbiologie Characteristics, Serologic Responses, and Clinical-Manifestations in Severe Acute Respiratory Syndrome, Taiwan 1391 Chest-X-Ray Imaging of Patients with SARs 1391 Severe Acute Respirator,' Syndrome (SARs) - The Questions Raised by the Management of a Patient in Besancon and Strasbourg 1391 Evaluation of Who Criteria for Identifying Patients with Severe Acute Respiratory Syndrome Out-of-Hospital - Prospective Observational Study 1391 Safe Tracheostomy for Patients with Severe Acute Respiratory Syndrome 1391 Description and Clinical Treatment of an Early Outbreak of Scvere-Acute- Respiratory-Syndrome (SARs) in Guangzhou. Pr- China 1391 Reovirus, Isolated from SARs Patients 1391 A Clinicopathological Study of 3 Cases of Severe Acute Respiratory Syndrome (SARs) 1391 Infection-Control for SARs in a Tertiary Neonatal Center 1391 Severe Acute Respiratory Syndrome in Haemodialysis-Patients - A Report of 2 Cases 1501 Clinical-Course and Management of SARs in Health-Care Workers in Toronto - A Case Series 1501 Outcomes and Prognostic-Factors in 267 Patients with Severe Acute Respiratory Syndrome in Hong-Kong 1501 Newly Discovered Coronavirus as the Primary Cause of Severe Acute Respiratory Syndrome 1501 Severe Acute Respiratory Syndrome in a Hemodialysis-Patient 1513 Severe Acute Respiratory-Distress-Syndrome (SARs) - A Critical-Care Perspective 1513 Enteric Involvement of Severe Acute Respiratory Syndrome- Associated Coronavirus Infection 1513 Investigation of a Nosocomial Outbreak of Severe Acute Respirator}' Syndrome (SARs) in Toronto, Canada 121 4.3 The Third Fusion Level - C3 Clusters 4.3.1 Clusters and Cluster Sizes The application of the complete link cluster method on the last level of cluster fusion resulted in a partition where numerous singleton clusters and a few clusters containing two objects were generated only. This means that an upper limit for the application of iterated clustering is found for the proposed method. Still, the question if links between clusters generated on the C2-level are able to form relevant clusters on the last level of cluster fusion needs to be answered. In order to be able to map such links, the between groups average cluster method was applied (see Sub-section 1.4.1 in Chapter 2). On basis of the computed AvgCS(C, C) between C2 clusters, 189 C2 clusters were partitioned into 92 singleton clusters and 38 clusters containing more than one C2 cluster (see Table 5-15). No threshold of AvgCS(C, C) was applied.48 The total sum of articles in the 38 clusters was 1,763. 48 Extremely low values were regarded as zero associations, hence, the large numbers of singleton clusters on the C3-level (See also Sub-sections 1.3.4 and 1.3.5 in Chapter 5). Table 5-15: The Size Frequency Distribution of C2-Clusters Interval Frequency 170-199 1 150-169 0 130-149 0 110-129 0 90-109 2 70-89 3 50-69 8 30-49 9 10-29 15 Sum 38 Note: Singleton clusters are excluded and the class interval is 20. The shape of the distribution is summarized in Figure 5-31. 122 Figure 5-31: The Distribution of Core Documents over C3-Clusters Fr eq ue nc y o> 12- 4— 6— 2- 8— 0-f— 0,00 50,00 100,00 150,00 200,00 Number of core documents in clusters uc As can be seen from Figure 5-31, the distribution is positively skewed, with most cluster sizes gathered at the lower range of the scale. The median cluster size was 37. Macro clusters (N > 86) are seen at the higher range of the scale and at the tail. The macro clusters are again from the field of physics (Condensed Matter, Particle Physics, Crystallography and Applied Physics) and from the bio-medical sciences (Oncology-Haematology). One physics cluster (Condensed Matter) builds partly on one of the macro clusters formed at the second fusion level, and one cluster from the bio-medical sciences (Oncology-Haematology) on two of the macro clusters formed at the second fusion level. Otherwise, macro clusters are generated by merging medium and smaller sized C2-clusters. 4.3.2 Coherence and Separation For the 38 C3-clusters, the AvgCS(C) was calculated. The shape of the distribution is summarized in Figure 5-32. 123 Figure 5-32: The Distribution of Coefficients of AvgCS(C) over 38 C3- Clusters Fr eq ue nc y 6- 4-: 8— 2- 0 .............. Illi "1......... 1,00 2,00 3,00 4,00 5,00 6,00 7,00 8,00 AvgCS(C) u 0) 0> As can be seen from Figure 5-32, the distribution is almost rectangular. Moving up to this level of cluster fusion, the mean AvgCS(C) drops from 7.95 to 3.64. Likewise, the density of links between core documents D makes a drop from near the maximum value to a mean of 0.66 (md = 0.64) (see Figure 5-33). Figure 5-33 : The Distribution of Coefficients of D over 38 C3-Clusters Fr eq ue nc y 12— 6— 9— 3- 0,50 0,60 0,80 0,90 1,000,70 D 0-f— 0,40 u □ o 124 With regard to the aspect of isolation, the AvgCS(C, C) between 38 C3- clusters was calculated. The shape of the distribution is summarized in Figure 5-34. Figure 5-34: The Distribution of Coefficients of AvgCS(C, C) over Links between 38 C3-Clusters 40— 20- 50— 30— 10- u c CD z CT CD 0,00 0,02 0,04 0,06 0,08 0,10 0,12 0,14 0,16 AvgCSfC, C) As can be seen from Figure 5-34, the distribution is positively skewed. Six C3-clusters are isolated. The median AvgCS(C, C) between the remaining 32 C3-clusters has dropped from 0.02 to 0.002 and mostly spurious links between C3-clusters remain. 4.3.3 Example of Cluster Fusion on the C3-Level On the final level of cluster fusion, cluster C2/170 formed on the second level of cluster fusion is merged with two other C2-clusters to C3/3, with a total number of 61 core documents (see Table 5-16). Table 5-16: The Fusion of Three C2-Clusters to C3/3 AvgCS(C, C’) No. Shared References C2- Cluster Cluster Size C2- Cluster Cluster Size 0.25 129 87 16 170 32 0.10 20 87 16 171 13 2.48 1030 170 32 171 13 As can be concluded from Table 5-16 above, cluster C2/87 presents the most distant (or dissimilar) vertex in a three-edged graph. Hence, from some aspect of similarity, it should deviate from the two other C2-clusters. This is, 125 however, not clearly reflected by the mix of journal subject categories assigned to the following core documents in the C2-clusters: C2/ 87: infectious diseases, clinical microbiology; C2/170: general & internal medicine, infectious diseases, biochemistry ; pediatrics; urology & nephrology; and C2/171 : biochemistry & molecular biology, clinical chemistry, microbiology, virology. Approximately 11 different disciplines are more or less associated with the research theme(s) of C3/3.49 All three C2 clusters constituting C3/3 were complete graphs, with regard to links between core documents. The values of Avg(C) ranged between 3.54 and 5.32. The values of D and AvgCS(C) for the graph of C3/3 were 0.62 and 2.30 respectively. Hence, the coherence was lower than for the average C3 cluster. 49 The exact number of disciplines will not be given by assigned journal subject categories as these are journal classifications, covering the scope of journals, not the scope of the individual paper. At this level of cluster fusion, the structure is more complex and worthy of a more thorough analysis. As such, a “top-bottom” interpretation is suggested. Beginning this analysis from the “top”, all links between core documents in C3/3 are displayed in a two dimensional plane by MDS with an acceptable value of stress (see Figure 5-35). In this graph, the constituent C2 clusters are clearly discernable. Clusters C2/171 and C2/170 are configured in the upper part of the map and cluster C2/87 in the lower part with the core document 400557 in an intermediate position. 126 Figure 5-35: The Configuration of C3/3 Constituted by C2/87, C2/170 & C2/171 73304/170 345210/171 28545/171 23546/171 28547/171 275300/171 603696/171 73303/170 3845/170 333937/171 179548/171 110241/170 91781/170 28525/170568577/170 2604/171 219805/171 275806/170 27218/170 335649/170 361876/170 423167/170 91791/170 27214/170 400553/170 288845/170 603695/171 , 27182/171 288844/170 333109/170 55760/170 27215/170 490401/170 8215/170 403574/170 219617/170 27178/170 400512/170 527101/170 383006/170 8202/170 431384/170 488422/170 488519/170 400557/87 268427/87 2326/87 179568/87 400623/87 289027/87 179849/87 130921/87 179840/87 566298/87 219926/87 400624/87 378605/87 209004/87 179739/87 566290/87 Note: i. Numbers on the map correspond to document numbers and C2-cluster numbers. ii. Kruskal's stress is 0.11 Applying MDS to display the associations between significant terms (title words) according to their co-occurrence in titles, an overview sketch of the topic content in cluster C3/3 is arrived at (see Figure 5-36). 127 Figure 5-36: MDS Display of the Co-occurrence of Title Words in the Core Documents of C3/3 :oronavlrus| SARs Padeats ‘Assays: koronto^ ‘Canadá 'OUHQ^Metapneumovkus^ Human^'^p§ ““•-Ac h tfdren US«-®''- : > w:®e'"* ,, . ÿ " ' - r /»Phytogeny, , C'.c ' Wx sx^Respiraäory Note: i. The lowest frequency of term occurrence allowed for is 2. ii. The width of links is corresponding to the strength of similarity between terms as measured by the Jaccard index (see Equation 2.4). iii. Circle sizes correspond to the frequency of occurrence. iv. Kruskal’s stress is 0.10. As can be seen from Figure 5-36, the focus of core documents in cluster C3/3 is on infectious diseases caused by viruses (in particular SARS). To begin with, the configuration in terms of term size-position could be examined. The term “SARs” has a central position and, as indicated by the circle size, it is the most frequent term. Radiating out from “SARs”, different dimensions associated with diseases caused by viruses can be discerned: i. a time-geography dimension (outbreak; early; epidemiology; Hong- Kong; China; Guangzhou; Taiwan; Canada; Toronto); ii. a clinical dimension (metapneumovirus; human; infection; children; young; infants; lower; respiratory tract; HIV 1; hospitalized; prevalence); and iii. a dimension of the genetics of viruses (analysis; genome; phylogeny; new; protein; genes; application; virus). 128 Zooming in on particular C2-clusters, different aspects of cluster C3/3 could be reflected. Beginning with cluster C2/87 (located on the lower part of the map in Figure 5-35), this cluster is constituted by two Cl clusters and the total number of core documents was 16. Studying the titles of core documents constituting C2/87, the subject homogeneity is obvious as they are all on Human Metapneumovirus. Also, a preliminary explanation of the intermediate role of core document 400557 (see Figure 5-35) is that it associates this virus with SARS. SARS is for all associated with a corona virus (SARS-CoV), but also with the human metapneumo virus, though to a lesser degree (see Table 5-17). Table 5-17: Core Document Titles in C2/87 2826 / Human Metapneumovirus-Associated Lower Respiratory-Tract Infections Among Hospitalized Human-Immunodeficiency-Virus Type-1 (HIV-1 )-lnfected and HIV-1-Uninfected African Infants 130921 / Prevalence and Clinical Symptoms of Human Metapneumovirus Infection in Hospitalized-Patients 179568 / Human Metapneumovirus Infection in the Canadian Population 179739 / Comparative-Evaluation of Real-Time PCR Assays for Detection of the Human Metapneumovirus 179840 / Human Metapneumovirus Associated with Respiratory-Tract Infections in a 3-Year Study of Nasal Swabs from Infants in Italy 179849 I High Prevalence of Human Metapneumovirus Infection in Young- Children and Genetic-Heterogeneity of the Viral Isolates 2090041 Human Metapneumovirus Infections in Young and Elderly Adults 219926 I Seroprevalence of Human Metapneumovirus in Japan 268427 I Effects of Human Metapneumovirus and Respiratory Syncytial Virus-Antigen Insertion in 2 3'-Proximal Genome Positions of Bovine /Human Parainfluenza Virus Type-3 on Virus-Replication and Immunogenicity 289027 I Human Metapneumovirus Infection in the United-States - Clinical-Manifestations Associated with a Newly Emerging Respiratory-Infection in Children 378605 I Human Metapneumovirus in a Hematopoietic Stem-Cell Transplant Recipient with Fatal Lower Respiratory-Tract Disease 400557 / Human Metapneumovirus Detection in Patients with Severe Acute Respiratory Syndrome 400623 / Children with Respiratory-Disease Associated with Metapneumovirus in Hong-Kong 400624 / Human Metapneumovirus Infections in Hospitalized Children 219926 / Seroprevalence of Human Metapneumovirus in Japan 566290 / Human Metapneumovirus Infection in Thai Children Clinical findings and studies of prevalence with regard to human metapneumo virus (with some emphasis on children) are presented by the core document titles. Looking at titles and journal subject category assignments of journals in which these core documents were published, the disciplinary structure leans towards general (internal) medicine (infectious diseases) but there is also a contribution from basic medical sciences (immunology, microbiology) (see Table 5-18). 129 Note: Numbers in brackets correspond to the frequency of articles published in ajournai. Table 5-18: Journal Titles and Assigned Journal Subject Categories Corresponding to Core Documents in C2/87 In C3/3 Journal Title Journal Subject Categories (1) Clinical Infectious Diseases Immunology; infectious diseases; microbiology (3) Emerging Infectious Diseases Immunology; infectious diseases (2) Journal of Infectious Diseases Infectious diseases (2) Scandinavian Journal of Infectious Diseases Infectious diseases (4) Journal of Clinical Microbiology Microbiology (1) Journal of Medical Virology Virology (1) Journal of Virology Virology (1) Pediatrics Paediatrics (1) Bone Marrow Transplantation Oncology; hematology; immunology; transplantation The next constituent C2-cluster to be studied is C2/171 (located on upper left quadrant in Figure 5-35) which is formed by three Cl clusters and 13 core documents. Table 5-19 gives the core document titles in this cluster. Table 5-19: Core Document Titles in C2/171 2604 / Quantitative-Analysis and Prognostic Implication of SARs Coronavirus RNA in the Plasma and Serum of Patients with Severe Acute Respiratory Syndrome 179548 / Evaluation of Reverse Transcription-PCR Assays for Rapid Diagnosis of Severe Acute Respiratory Syndrome-Associated with a Novel Coronavirus 219805 / Early Events of SARs Coronavirus Infection in Vero Cells 27182 / Establishment of a Fluorescent Polymerase-Chain-Reaction Method for the Detection of the SARs-Associated Coronavirus and Its Clinical-Application 28545 / Design and Application of 60Mer Oligonucleotide Microarray in SARs Coronavirus Detection 28546 I Molecular Phylogeny of Coronaviruses Including Human SARs-Cov 285471 Phylogeny of SARs-Cov as Inferred from Complete Genome Comparison 333937 / Activation of Ap-1 Signal-Transduction Pathway by SARs Coronavirus Nucleocapsid Protein 345210 / Genomic Characterization of the Severe-Acute-Respiratory- Syndrome Coronavirus of Amoy Gardens Outbreak in Hong-Kong 603695 / Coronavirus in Severe Acute Respiratory Syndrome (SARs) 603696 I Severe Acute Respiratory Syndrome - Identification of the Etiologic Agent 62784 / Mutation Analysis of 20 SARs Virus Genome Sequences - Evidence for Negative Selection in Replicase Orf1B and Spike Gene 275300 I The Crystal-Structures of Severe Acute Respiratory Syndrome Virus Main Protease and Its Complex with an Inhibitor As can be seen from Table 5-19, this cluster focuses exclusively on corona virus and SARS. The emphasis is on the analysis of the genetic 130 characterization and on methods for the detection and description of the viruses (e.g. laboratory methods, isolation and cultivation). The clinical focus seen in cluster C2/87 is thus replaced with more basic research in the virus causing SARS. This is also reflected by the composition of the set of publishing journals and journal subject categories of this cluster, where the contribution from chemistry and biochemistry is salient (see Table 5-20). Note: Numbers in brackets correspond to the frequency of papers published in ajournai. Table 5-20: Journal Titles and Assigned Journal Subject Categories Corresponding to Core Documents in Cluster C2/171 Journal Title Journal Subject Categories (3) Chinese Science Bulletin Multidisciplinary sciences (2) Trends in Molecular Medicine Biochemistry & molecular biology; cell biology; medicine, research & experimental (1) ACTA Pharmacologica Sinica Chemistry, multidisciplinary; pharmacology & pharmacy (1) Biochemical and Biophysical Research Communications Biochemistry & molecular biology; biophysics (1) Chinese Medical Journal Medicine, general & internal (1) Clinical Chemistry Medical laboratory technology (1) Journal of Clinical Microbiology Microbiology (1) Journal of Medical Virology Virology (1) Lancet Medicine, general & internal (1) Proceedings of the National Academy of Sciences of the United States of America Multidisciplinary sciences Lastly, the largest C2 cluster (i.e. C2/170) constituting C3/3 was studied (located on upper right quadrant in Figure 5-35). This cluster was formed by three Cl clusters and 32 core documents. In this cluster, several case studies as well as clinical aspects on diagnosis and prevention are reported, and the overall focus is again on clinical aspects of SARS (see Table 5-21 ). 131 Table 5-21: Core Document Titles in Cluster C2/170 3845 / Severe Acute Respiratory-Distress-Syndrome (SARs) - A Critical-Care Perspective 8202 / Investigation of a Nosocomial Outbreak of Severe Acute Respiratory Syndrome (SARs) in Toronto, Canada 8215 I Clinical-Course and Management of SARs in Health-Care Workers in Toronto - A Case Series 27178 / Chest-X-Ray Imaging of Patients with SARs 27214 / A Hospital Outbreak of Severe Acute Respiratory Syndrome in Guangzhou, China 27215 / Clinical Analysis of 45 Patients with Severe Acute Respiratory Syndrome 27218 / The Role of Radiological Imaging in Diagnosis and Treatment of Severe Acute Respiratory Syndrome 28525 / Reovirus, Isolated from SARs Patients 55760 / Initial Otolaryngological Manifestations of Severe Acute Respiratory Syndrome in Taiwan 73303 / Severe Acute Respiratory Syndrome in a Hemodialysis-Patient 73304 I Clinical Presentation and Outcome of Severe Acute Respiratory Syndrome in Dialysis Patients 91781 / Outcomes and Prognostic-Factors in 267 Patients with Severe Acute Respiratory Syndrome in Hong-Kong 91791 / An Outbreak of Severe Acute Respiratory Syndrome Among Hospital Workers in a Community-Hospital in Hong-Kong 110241 / Infection-Control for SARs in a Tertiary Neonatal Center 219617 / Description and Clinical Treatment of an Early Outbreak of Severe-Acute- Respiratory-Syndrome (SARs) in Guangzhou, Pr- China 275806 / A Clinicopathological Study of 3 Cases of Severe Acute Respiratory Syndrome (SARs) 288844 / A Young Infant with Severe Acute Respiratory Syndrome 288845 / Children Hospitalized with Severe Acute Respiratory Syndrome- Related Illness in T oronto 333109 I Severe Acute Respiratory Syndrome (SARs) - The Questions Raised by the Management of a Patient in Besancon and Strasbourg 335649 I Maintaining Dental Education and Specialist Dental-Care During an Outbreak of a New Coronavirus Infection - Part 1 - A Deadly Viral Epidemic Begins 361876 I Zcurve-Cov - A New System to Recognize Protein-Coding Genes in Coronavirus Genomes, and Its Applications in Analyzing SARs-Cov Genomes 383006 I Evaluation of WHO Criteria for Identifying Patients with Severe Acute Respiratory Syndrome Out-of-Hospital - Prospective Observational Study 400512 / Severe Acute Respiratory Syndrome-Associated Coronavirus Infection 400553 I Role of China in the Quest to Define and Control Severe- Acute-Respiratory- Syndrome 400574 I Microbiologie Characteristics, Serologic Responses, and Clinical-Manifestations in Severe Acute Respiratory Syndrome, Taiwan 423167 I Prediction of Proteinase Cleavage Sites in Polyproteins of Coronaviruses and Its Applications in Analyzing SARs-Cov Genomes 431384 I Enteric Involvement of Severe Acute Respiratory Syndrome- Associated Coronavirus Infection 488422 / Epidemiology and Cause of Severe Acute Respiratory Syndrome (SARs) in Guangdong, Peoples-Republic-of-China, in February, 2003 488519 / Newly Discovered Coronavirus as the Primary Cause of Severe Acute Respiratory Syndrome 490401 / Safe Tracheostomy for Patients with Severe Acute Respiratory Syndrome 527101 / Severe Acute Respiratory Syndrome in Hemodialysis-Patients - A Report of 2 Cases 568577 / Transmission Dynamics of the Etiologic Agent of SARs in Hong- Kong - Impact of Public-Health Interventions 132 Looking at the distribution of journal subject categories and journal titles, several medical disciplines and sub-disciplines are represented with an emphasis on general medicine (see Table 5-22). Note: Numbers in brackets correspond to the frequency of papers published in ajournai. Table 5-22: Journal Titles and Assigned Journal Subject Categories Corresponding to Core Documents in Cluster C2/170 Journal Title Journal Subject Categories (5) Chinese Medical Journal Medicine, general & internal (3) Emerging Infectious Diseases Immunology; infectious diseases (2) American Journal of Kidney Diseases Urology & nephrology (2) Annals of Internal Medicine Medicine, general & internal (2) Canadian Medical Association Journal Medicine, general & internal (2) Lancet Medicine, general & internal (2) Paediatrics Paediatrics ( 1) Archives of Disease in Childhood Paediatrics (1) Archives of Otolaryngology-Head & Neck Surgery Otorhinolaryngology; surgery (1) British Dental Journal Dentistry, oral surgery & medicine (1) British Medical Journal Medicine, general & internal (1) Chinese Science Bulletin Multidisciplinary sciences (1) Critical Care Medicine Critical care medicine (1) FEBS Letters Biochemistry & molecular biology; biophysics; cell biology (1) Gastroenterology Gastroenterology & hepatology (1) Journal of Medical Microbiology Microbiology (1) Laryngoscope Medicine, research & experimental; otorhinolaryngology (1) Nephrology Dialysis T ransplantation Transplantation; urology & nephrology (1) Pathology Pathology (1) Presse Medicale Medicine, general & internal (1) Science Multidisciplinary sciences It is to be noted that the different MDS maps (i.e. Figure 5-35 and Figure 5-36) match surprisingly well, though the first map groups papers according to shared references and the second map title words according to their co­ occurrence frequency in titles in core documents. Hence, C2/87 seems to correspond to the lower part of the “title word” map, C2/171 to the left­ middle-upper part and C2/170 to the right-middle-upper part. It seems clear that the C2-clusters are subject consistent and a common denominator (SARS) can be identified. The interdisciplinary character on all levels (Cl to C3) is obvious. The merging of the three C2-clusters on the C3 level thus connects research in two different viruses and the pathology of 133 diseases caused by them, genesis of agents and corresponding clinical research and observations. The merging of C2/87 with the other two clusters could, however, be questioned. Though a least common denominator (SARS) exists, both the measured distances and the assessed cognitive distances between C2- clusters showed that there is a need for the separation of cluster C2/87 from the other C2-clusters. Nevertheless, the associations of the clusters could be of interest as they all gather around a common problem, though from different perspectives. 4.4 Field Experts’ Evaluations of Four Cases of Iterated Clustering Four cases of iterated clustering were presented to four different field experts for evaluation. The design of this experiment aimed at finding clusters on the last level of cluster fusion from the three major science fields: physics, chemistry and bio-medicine for the evaluation. C3-clusters from these fields were then matched against profiles of researchers and when a match occurred, a preliminary choice of cluster was made. If a researcher was available to do the evaluation, a final choice of cluster was made. In order to approximate the greater impact of physics on the composition of the database underlying the experiments in Case 4, two cases from the field of physics, and one each from the other two fields were selected. In all. a total of 154 core documents were evaluated. Corresponding to each case is a C3-cluster, which is assumed to reflect research themes with a cognitive linkage to one another. The field experts were asked to assess the relevance of cluster composition on all three levels of cluster fusion, i.e. Cl to C3 (see Sub-section 2.1 in Chapter 4). In order to illustrate the field experts’ evaluations, the internal structure of each C3- cluster is visualized by mapping links between constituent core documents. It is to be noted that there exists a scale difference between maps and they are not directly comparable with each other. The results of the evaluations are given below. 4.4.1 Cluster C3/12, “Human Genetics and Disease” This C3-cluster contained 53 core documents distributed over three C2- clusters and 11 Cl-clusters as follows: C2/45 : CI/616; CI/1003 C2/46 : CI/1171; CI/1297; CI/1170; CI/1172 C2/210: Cl/9; Cl/1163; Cl/1168; CI/1184; CI/1286 The focus in this cluster is generally on human genetics and disease. A total of 21 different journal subject categories were assigned the journals in which core documents in this cluster were published (see Table 5-23).50 50 The exact number of contributing disciplines will not be given by assigned journal subject categories as these are journal classifications, covering the scope of journals, not the scope of the individual article. 134 Table 5-23: The Frequency Distribution of Journal Subject Categories in Cluster C3/12 Frequency Journal Subject Categories 22 Genetics & Heredity 13 Gastroenterology & Hepatology 5 Dermatology 4 Immunology 3 Biochemical Research Methods 3 Biochemistry & Molecular Biology 3 Biotechnology & Applied Microbiology 3 Cell Biology 3 Computer Science, interdisciplinary applications 3 Mathematics, interdisciplinary applications 3 Multidisciplinary Sciences 3 Statistics & Probability 2 Pathology 1 Nutrition & Dietetics 1 Ophthalmology 1 Pediatrics 1 Pharmacology & Pharmacy 1 Public, Environmental & Occupational Health 1 Respiratory System 1 Rheumatology In the graph representing C3/12, each of the three C2-clusters are clearly depicted and demarcated as can be seen from Figure 5-37. The density D of the graph was 0.52 and the AvgCS(C) 2.85. Hence, both values of cluster coherence were below the average. 135 Figure 5-37: The Configuration of Core Documents in Cluster C3/12 ¡12285 C2/210 ¡17454 783229 C2/45 '1043^ •10161' 14141: *420921; ?7?1 ■14122O7| 'f^^p1843$ ^41220WW72i >9107*«rf>2298 '2855 gu F72850 r " C2/4636/425 ; z’75516:/ ,«* ‘¿^ 1 i ’39395 »257195 Note: Kruskal’s stress is 0.06. The field expert’s opinion was that core documents in Cl and C2 clusters were consistently subject related, with two exceptions. The first exception is article 466464 in C2/210/C 1/1184 which seemed to have a too general topic content in relation to the pronounced focus in C2/210 on inflammatory bowel-diseases. This was in agreement with its more peripheral position on the map. The second exception was core document 283229 in C2/210/C1/1163, which the expert assumed to be relevant, but with some uncertainty as the title was not exhaustive enough. The field expert renounced judging the relevance of merging disease-gene-mapping methods (C2/46), with research in genetic aspects of psoriasis (C2/45) and inflammatory bowel-disease (C2/210). 4.4.2 Cluster C3/19: “Chemistry” This cluster contained 25 core documents distributed over two C2-clusters and five Cl-clusters as follows: C2/111: CI/1189; CI/1263 C2/191: CI/81; CI/406; CI/1394 136 All core documents but one in C3/19 pertained to the field of chemistry and the composition of contributing disciplines varied, though with an emphasis on organic chemistry. A total of five different journal subject categories were assigned the journals in which core documents in this cluster were published (see Table 5-24).51 51 The exact number of contributing disciplines will not be given by assigned journal subject categories as these are journal classifications, covering the scope of journals, not the scope of the individual article. Table 5-24: The Frequency Distribution of Journal Subject Categories in Cluster C3/27 Frequency Journal Subject Categories 17 Chemistry, Organic 6 Chemistry, Multidisciplinary 1 Chemistry, Inorganic & Nuclear 1 Cell Biology 1 Chemistry, Applied In the graph depicting C3/19, the composition of C2/191 is visualized as a compact cluster whereas C2/111 is a looser construct, and there exists no links between CI/1263 and C2/111 (see Figure 5-38). In spite of the latter, the coherence of C3/19 is above the average with 5.32 for the AvgCS(C) and 0.67 for D. 137 Figure 5-38: The Configuration of Core Documents in Cluster C3/19 '592012C2/191 292747 >723 C2/111 C1/1263 >36949 50621 124118 587193 591502 591399 174408 448569 26957 443265 à 't- V 'iA ft ■» ft ft ft I ft ft ft C1/1189 A----------- A 591564 «Æ®0* 5<)^jteis9 Note: i. Due to the compactness of C2/I91, seven labels representing articles could not be fitted to mark corresponding circles of cluster C2/191 and are presented in the nearby table in the map. ii. Kruskal’s stress was 0.02. According to the field expert, no misplaced core documents were found on the Cl-level. However, C2/11 I/CI/1263 was found to be more diverse than cluster C2/111/Cl/l 189, which is reflected by the configuration in the map where Cl/1263 form a looser structure. On the C2 level, C2/111 was considered to be subject consistent in terms of a research focus common to the constituent Cl clusters. As for C2/191, the partition in Cl-clusters appeared artificial to the field expert and C2/191 was better regarded as one cluster, which is reflected by this cluster’s compactness, as seen in the map. Regarding the merging of the C2 clusters, no clear subject relationship between them was obvious. 138 4.4.3 Cluster C3/27: “Bose-Einstein Condensation” This cluster contained 54 core documents distributed over four C2-clusters and 13 Cl-clusters as follows: C2/140: CI/367; CI/555 C2/141: CI/459; CI/557; CI/578 C2/143: CI/353; CI/362; CI/439; CI/552 C2/144: CI/352; CI/359; CI/454; CI/643 Articles in this cluster pertain to research areas of optical, atomic & molecular physics and the major focus is on Bose-Einstein condensation.52 A total of five different journal subject categories were assigned the journals in which core documents in this cluster were published (see Table 5-25).53 52 Bose-Einstein condensation is the collapse of atoms into a single quantum state. 53 The exact number of contributing disciplines will not be given by assigned journal subject categories as these are journal classifications, covering the scope of journals, not the scope of the individual article. Table 5-25: The Frequency Distribution of Journal Subject Categories in Cluster C3/27 Frequency Journal Subject Categories 19 Optics 19 Physics, Atomic, Molecular & Chemical 16 Physics, Multidisciplinary 3 Physics, Condensed matter 1 Multidisciplinary sciences 1 Physics, Applied In the graph depicting C3/27, each of the four C2 clusters is clearly demarcated (see Figure 5-39). The density D is 0.43 and the AvgCS(C) 1.97, hence both values are clearly below the average. 139 Figure 5-39: The Configuration of Core Documents in Cluster C3/27 !286i C2/143 76886E 0008. '2982?. X;»7308S62 im ‘W -c? V ÿg V99 ”4 . . ïJÇÎyXH. XV*-3$SLk vf ’“P >44146» ¡V' ‘ fl iS^l I,.../ -< SSfei.., <■■- " 3869Ç ' £ m ’S0Í469 e!q-c fe SSSzàââ--”-- j~--" ’::•:•••■• ' ••*<:?::•:•••• :■:•?>>•< 4-.:: :':<•■• .Is* K WÍA ■ ^^539^.29£1,, M^24¿-~4 g S .... -«*^4-!74261 11S13100 10183 . Note: Kruskal’s stress is 0.08. The field expert presented in this case an elaborated evaluation where not only misplaced core documents on the Cl-level were considered, but also minor deviations between their research foci. The remarks given by the field expert with regard to the cluster composition at the Cl level are as follows: i. In cluster C2/140/ CI/367, core document 308862 caused some uncertainty as the subject content as reflected by its title was not completely transparent. ii. In cluster C2/140/C1/155, core document 299154 had a somewhat deviating focus in comparison with other cluster members. iii. In cluster C2/143/C1/353, core document 298656 had a slightly deviating focus in comparison with other cluster members. Also, core document 308469 and core document 308699 cohered, but were 140 considered somewhat deviating in relation to core document 277841 and core document 308699, which formed a coherent pair. Hence, this cluster “sprawled” slightly in terms of cluster coherence. iv. In cluster C2/143/C1/362, core document 297989 deviated somewhat from the other core documents in C3/27. v. No core document deviated to the extent that it should be considered as clearly misplaced. vi. Concerning C2-clusters, the field expert’s opinion was that all constituent Cl-clusters shared the same research focus, hence, all C2- clusters belonged to the same area of research. Conclusively, some deviations on the Cl level were detected and when core documents were aggregated to higher levels, a common research theme for all core documents in C3/27 is seen. 4.4,4 Cluster C3/29: “Carbon-Nano-Tubes” This cluster contained 22 core documents distributed over two C2-clusters and five Cl-clusters: C2/27: CI/1018; CI/1416 C2/28: CI/549; CI/1072; CI/1137 Core documents in C3/29 focus on carbon-nano-tubes (CNTs) from different angles.54 A total of 11 different journal subject categories were assigned the journals in which core documents in this cluster were published (see Table 5- 26).55 54 Carbon nano tubes are cylindrical carbon molecules with properties that make them potentially useful in extremely small scale electronic and mechanical applications. They exhibit unusual strength and unique electrical properties, and are efficient conductors of heat. 55 The exact number of contributing disciplines will not be given by assigned journal subject categories as these are journal classifications, covering the scope of journals, not the scope of the individual article. 141 Table 5-26: The Frequency Distribution of Journal Subject Categories in Cluster C3/27 Frequency Journal Subject Categories 5 Chemistry, analytical 5 Physics, applied 4 Physics, condensed matter 3 Chemistry, physical 3 Materials Science, multidisciplinary 2 Physics, atomic, molecular & chemical 1 Biochemical Research Methods 1 Engineering, electrical & electronic 1 Multidisciplinary sciences 1 Physics, multidisciplinary 1 Polymer science In the graph depicting C3/29, a complex and less clear cluster structure is reflected. Hence, the division of the map in C2-clusters and the subdivision in Cl-clusters are not clearly mirrored by the configuration of the graph representing C3/29 (see Figure 5-40). The density D was 0.70 and the AvgCS(C) was 2.45. Hence, the general level of interconnectedness is above the average, but the average strength of links is below the average. 142 Figure 5-40: The Configuration of Articles in Cluster C3/29 CI/549 & CI/1072 3166‘ C2/28 C2/27 Î00759 !2248 66722 72522 '586810 *24658 .88532^ ' S...... "'S38825 ZSöFSfe17024 ¡568685 AN » IT: 1 ¿í -fy) *—¿di X¿246548 CI/10 18 •— Note: i. The angled line dividing the map indicates the border between cluster C2/27 and cluster C2/28. ii. Kruskal’s stress was 0.07. The more complicated structure was also reflected in the field expert’s evaluation. To begin with, cluster Cl/1416 contained one misplaced core document (500759) as did cluster CI/549 (core document 246581). Moving to the C2 level, in C2/27, both CI/1018 and CI/1416 handled CNT growth, though CI/1018 focused on the growth of aligned CNT on patterned substrates whereas CI/1416 was about non-aligned growth. In C2/28 (containing CI/549, CI/1072 and CI/1137), Cl-clusters focus on CNTs from divergent perspectives with no obvious common theme which would justify their fusion to a C2 cluster. Concerning the subject relationship between cluster C2/27 and cluster C2/28, all Cl-clusters explicitly focused on CNTs except for CI/1137 where the interest in CNTs was deemed secondary. 56 However, the field expert renounced the evaluation of the relevance of merging the C2-clusters. 56 Cluster C2/1137 focused primarily on film-electrodes though all but one core document title had the term “carbon nanotubes” in the title. 143 4.5 The Expansion of Cl-Clusters In order to examine the extent to which the proposed method gives rise to a fragmentation of research specialties when applied for core document mapping, a complete mapping of significant links connecting core documents in a Cl-cluster with core documents extrinsic to the Cl-cluster is needed. Computing all such links with a NCS > 0.25, the ability of Cl clusters to expand was assessed. In this experiment, the expansion of clusters was tried on all clusters with a size > 1.57 It was found that on the average, a cluster could expand by eight times its size, consequently 12.5 percent of the articles in an expanded cluster typically constituted the original cluster58 (see Figure 5- 41). 57This means links between core documents in Cl-clusters with a size > 1 and all other 5,771 core documents from the first level of cluster fusion. 58 Let N be the number of documents in the original cluster. Let T be the number of documents in the expanded cluster: N ' 8= T, N/T= 0.125 and 8'0.125= I (100%). Figure 5-41: The Distribution Of Shares Of Original Clusters In Expanded Clusters Fr eq ue nc y 250— 200— 300— 100- 150- 50— 0' I I | I I I I | I I fl | ----------------1- 0,10 0,20 0,30 0,40 0,50 0,60 0,70 u 0)□ 3. On each fusion level, a share of clusters that did not fulfill the requirements for cluster fusion emerged. This way, by each level of cluster fusion, Cl to C3, the original set of core documents was reduced as the sizes of clusters increased. The stepwise loss of core documents and simultaneous increase in cluster size is presented in Table 5-29. Table 5-29: Three Levels of Cluster Fusion: Effects on Document Populations, Frequency of Clusters and Cluster Sizes Level of Fusion No. of Clustered Core Documents No. Of Clusters Median Cluster Size C1 4477 1,000 4 C2 3,524 212 24 C3 1,763 38 37 Note: i. The calculation of median cluster size does not include singleton clusters. ii. On the Cl level, clusters have a minimal size of three articles. iii. On the C2- and C3- levels, clusters are composed by at least two objects (clusters from earlier fusion levels). Concerning the aspect of external cluster isolation, by each level of fusion, the share of isolated clusters was increased while the strength of association between clusters was weakened. At the same time, the internal cluster coherence was weakened too. Comparing levels, the most drastic change with regard to the separation between clusters take place when moving up to the C2 level, while the most drastic change with regard to cluster coherence takes place when moving up to the C3-level (see Table 5-30). L Table 5-30: Three Levels of Cluster Fusion: Effects on Cluster Coherence and Cluster Isolation Level of Fusion C1 Percentage Isolated Clusters 5 Median AvgCS(C, C) 0.65 Mean AvgCS(C) 10 .58 Median D 1 00 C2 11 0.02 7.95 1.00 C3 16 0.00 3.64 0.64 149 With regard to statistical data, the optimal level of cluster fusion should be the C2-level. The reasons are as follows: i. On the Cl-level, associations between core documents in different clusters are strong. ii. On the C2-level, clusters are generally still coherent and well separated. iii. On the C3-level, clusters are considerably less coherent. These findings should be related to the field experts’ evaluations of the four C3-clusters. On the Cl-level, clusters were generally considered subject coherent. On the C2 level, in one case (C3/29), one C2-cluster was considered artificial. On the C3-level, only two of four C3-clusters could be evaluated with regard to the merging of C2-clusters, and one of these was considered irrelevant. Hence, field experts' evaluations did not contradict statistical findings. Lastly, though findings regarding iterated clustering of core documents indicated the breaking up of specialties by the generation of coherent C2- clusters, they did not cover for all associations between core documents in a Cl-cluster and core documents extrinsic to it as the partition in clusters itself breaks up links. Hence, mapping all links between core documents in a cluster and core documents extrinsic to the cluster with a minimal NCS of 0.25, it was clearly shown that core document clusters on the Cl-level constitute fragments of larger research themes. 150 CHAPTER 6: DISCUSSION AND CONCLUSIONS This chapter wraps up the study undertaken. It begins with a discussion of the empirical findings which are summarized and discussed. The last section gives the conclusions drawn on this study. 1. DISCUSSION 1.1 Cases 1 to 3 1.1.1 The Relevance of Clusters Generated by the Complete Link Cluster Method A small research field would generally imply weaker and fewer links of bibliographic coupling as a result of a lower publication output and a smaller base literature. Hence, methods of bibliographic coupling would generally be less applicable on research fields with a lower publication output (Glänzel & Czerwon, 1996). Furthermore, methods of bibliographic coupling should be applied with quite severe thresholds of NCS in order to secure significant associations between articles (Sen & Gan, 1983; Glänzel & Czerwon, 1996). However, with regard to Case 1 and Case 3, it was shown that the proposed method is capable of generating relevant clusters also on low levels of NCS (see Sub-section 3.1 in Chapter 4). In Case 2, the larger population of articles from the field of Organic Chemistry was applied as the test arena, facilitating the application of considerably more severe thresholds, and also the identification of a delimited set of core documents. This was also reflected by higher values of AvgCS(C) and in a comparably lower share of misplaced articles (2 percent). The microanalysis in Case 2 illustrated the ability of the method when applied on a single but large research field to map core documents and generate relevant clusters as no article was regarded as misplaced.59 The high relevance of generated clusters in the microanalysis of core documents was underlined by the agreement between results accomplished by the complete link clustering and MDS. 1.1.2 The Extent and Nature of Deviations Between Results Generated by the Complete Link Cluster Method and Results Generated by Intellectual- Manual Partitions It has been shown that in all cases there were large differences between the intellectual-manual clusterings performed by the field experts and the complete link clusterings. To begin with, the distribution of articles over clusters deviated in the sense that the partitions generated by the complete link cluster method resulted in more and smaller clusters and a lesser concentration of articles to clusters. This difference was explicitly illustrated by Pratt's measure of concentration (see Table 6-1). 50 One article was not evaluated though. 151 Table 6-1: The Concentration of Articles to Clusters in Cases 1 to 3 Case Complete Link Clusters Experts’ Clusters No. Of Clusters Pratt’s Measure No. Of Clusters Pratt’s Measure 1 17 0.13 10 0.39 2 44 0.17 17 0.57 3 40 0.07 36 0.42 With regard to internal coherence, the field experts’ clusters were generally less coherent with respect to both the strength and the density of links (see Table 6-2). Table 6-2: The Internal Coherence of Clusters in Case 1 to 3 Case Complete Link Clusters Experts’ Clusters Md AvgCS(C) Md AvgCS(C) D 1 3.67 1.90 0.61 2 13.94 3.91 0.30 3 4.17 1.65 0.33 Note: i. Clusters generated by the complete link cluster method have a default maximal value of 0(1.0). ii. Singleton clusters generated by the field experts are excluded in the calculations. With regard to the aspect of external isolation, clusters generated by field experts were generally less isolated as reflected by the shares of all 2- combinations of clusters that were coupled. With regard to the AvgCS(C, C), but with the exception for Case 1, the association between clusters were weaker when the complete link cluster method was applied. However, differences were not pronounced (see Table 6-3).60 60 The complex interplay between measures of coherence and isolation must be interpreted on the level of a particular case. This may provide the researcher with a detailed understanding of a field’s cognitive structure and the impact of chosen methods. Here, the aim of analysis is delimited to study deviations between partitions generated by two different methods. 152 Table 6-3: The External Isolation of Clusters in Case 1 to 3 Case Complete Link Clusters Experts’ Clusters A B C A B C 1 0.19 0.42 0 0.16 0.78 0 2 0.24 0.12 5 0.40 0.35 4 3 0.17 0.07 5 0.36 0.13 0 Note: A-columns contain median coefficients of AvgCS(C, C). The median was calculated excluding any isolated clusters that occurred; B-columns contain the shares of 2-combinations of clusters that were coupled and C-columns contain the number of isolated clusters. On a detailed level, the agreements between partitions were assessed by tabulating distributions by different sort orders, facilitating an exact visualization of deviations. Generally, little agreement between partitions was seen and some extreme deviations were found. For all, partitions generated by the complete link cluster method resulted in a more fine graded division, and often in a split up of experts' clusters, which was also illustrated by MDS. 1.1.3 A Commentary on and Comparison of Methods of Partition Contrasting the relatively high degree of relevance in clusters generated by the complete link cluster method with the pronounced deviations between the partitions generated by this cluster method and experts’ partitions, one may assume that there exist alternative classifications for Cases 1 to 3. Below are some suggestions that are in line with this assumption. The complete link cluster method applies common references as a measure of similarity exclusively. The intellectual clustering, on the other hand, was foremost based on semantic relations between titles (and abstracts) in different articles and largely independent of common references. Semantic relations between articles may well exist also when citation relations are weak or absent, as may citation relations between articles exist when there is an unclear or absent semantic relation. The latter would, for instance, be the case when different specialties are merged into new interdisciplinary areas and different terms are used to denote the same objects or phenomena. Hence, the two methods of partition should generate similar results only if semantic relations and citation relations converge. Moreover, the classification of articles accomplished by the complete link cluster method is dependent on how current research proceeds and undergoes changes as reflected by the use (referencing) of previous research. Due to the dynamic aspect of research, classifications (clusters) are not easy to anticipate. An associated issue is the demarcation of research themes with regard to the choice of hierarchical level for a cluster solution. As there exists no common framework for the demarcation of a field’s division in specialties, the delineation of borders between specialties or disciplines may well provide 153 difficulties.61 These aspects were to some extent reflected in some of the field experts’ comments, which indicated that more than one cluster solution may be acceptable. In Case 1, the field expert admitted the split up of some expert clusters when compared with the clusters generated by the complete link cluster method. In Case 2, the field expert noted that a new set of principles, hard to anticipate on beforehand, for partitioning and classification emerged when studying clusters generated by the complete link cluster method. In Case 3, the field expert concluded that the classification of articles could have different points of departure, and that a more fine graded partition of articles as well as the merging of some groups may be equally valid. 61 As stated in Tijssen (1992, p. 31 ), “[u]nlike geographical maps, maps of science are not directly related to the physical world” as there exists no common frame of reference. 1.4 The Effects of Threshold Settings and Method of Partition on the Original Populations of Research Articles As discussed in Sub-section 3.3 in Chapter 2, the issue of the extent to which topics covered by a population of research articles are identified by the applied method should be of interest. In this study the “recall” of relevant articles can not be directly assessed as the number of articles that have a cognitive (semantic) relation to a certain cluster is not known. However, assessing the effects of applied thresholds and cluster method on the sizes of the original populations of research articles, a coarse estimate of the proposed method’s ability to intercept current research themes of the populations under study could be provided. As selection criteria always will lead to a reduction of a population of articles, the diminishing of the sizes of the original populations of articles is granted. Several factors affect the extent to which an original population will be diminished. The more important factors are: i. the extent of consensual referencing of the field under investigation; ii. the set threshold of coupling strength or NCS ; and iii. the choice of cluster method. Concerning (i), this should be the most important factor deciding the extent to which research themes of the original population of articles is covered. Hence, when similar topics are treated but the referencing is less consensual (or attentive), articles will be lost. The share of articles not belonging to any cluster of the set minimal size is thus a reflection of the extent of non- consensual referencing for a particular population. With regard to (ii), there exists no straight forward method for deciding the most appropriate threshold, hence empirical experience would guide decisions of thresholds and methods may initially be more or less arbitrary (cf. Sub­ section 3.3 in Chapter 2). Generally, a large research field where specialties 154 have a clear and consensual focus would provide stronger links of bibliographic coupling and more choices of threshold setting. With regard to (iii), the choice of cluster method has an impact on the sizes of clusters. Generally, the more severe conditions to fulfill for the merging of articles to clusters, the greater the number of smaller sized clusters and the subsequent loss of articles when excluding clusters below a stipulated threshold of cluster size. It is clear that points (i) to (iii) are interrelated and that this interrelationship is complex and difficult to foresee. The empirical findings in Cases 1 to 3, reflect the impact of these factors on the original sizes of populations, which is illustrated in Table 6-4. Table 6-4. The Successive Diminishing of the Original Populations. Case Original Size A B 1 232 185 63 2 14,389 268 183 3 879 579 130 Note: Column A shows the sizes of the populations after threshold setting of the coupling Strength or NCS (Case 2) and column B shows the sizes of the populations when clusters containing less than three articles have been excluded. Starting with Case 1, the original size of the population of articles was 232. The method of noise reduction (the application of a threshold of one coupling unit) implied a further diminishing of the set by approximately 20 percent to a total of 185 articles. After clustering and exclusion of clusters where the size < 3, only 34 percent of these articles remained. Conclusively, the total reduction was 73 percent. In Case 2, the filtering out of bibliographically coupled pairs with a NCS below 0.25 from the remaining articles and applying a threshold of four links at the same threshold of NCS brought about a reduction of approximately 98 percent. Considering the effect of applying cluster size threshold, the total reduction of articles was 99 percent. In Case 3, with regard to the noise reducing actions taken, the same approach of threshold setting as in Case 1 was applied (one coupling unit). This brought about a diminishing of the set of articles by 34 percent to a final set of 579 articles. This set was further reduced by 78 percent to 130 articles by excluding clusters containing less than three articles. Hence a total reduction by 85 percent. It can be concluded that even if low thresholds (Cases 1 and 3) of coupling strength are applied, a notable reduction of the sizes of the original populations of articles takes place. 155 1.1.5 Implications of Findings Only a small fraction of articles in the original populations was included in the mappings, most likely implying the absence of research themes as well as articles potentially relevant for the subject foci of clusters. Generally, there exists a clash between relevance and interception as severe thresholds imply a considerable loss of articles, whereas the absence or application of low thresholds may impair the relevance of clusters. A similar clash should exist with regard to the choice of cluster method as a method with severe conditions to fulfill (e.g. the complete link cluster method) would bring about an increased loss of articles in comparison with a more generous method (e.g. the single link cluster method), but promote relevance. When the prime objective is to find relevant information, a gearing of thresholds may tentatively be applied for information provision purposes. As was shown, science fields of different sizes and referencing characters may be mapped by the applied method and when the network of bibliographically coupled articles of a selected population so allows for, thresholds may be varied so that both cores of consensual research as well as more comprehensive but perhaps less significant cluster structures are identified. With regard to the latter, most probably, the application of lower thresholds in Case 2 would still generate useful but perhaps less lucid information. The fact that partitions generated by field experts consequently and strongly deviated from partitions generated by the complete link cluster method means that the proposed method did not converge with field experts’ comprehensions of fields’ cognitive structures. Hence, if accepting field experts' apprehensions of scientific structures as valid points of reference, the findings indicated that the proposed method may not clearly identify conceptualized structures, hence its capability of laying out the cognitive structures of specialties should not only on theoretical grounds (see Sub-section 3.2 in Chapter 2) be ambiguous. However, subject coherent clusters containing relevant information were generated over all three cases and the method was capable of identifying smaller, coherent research foci on comparably low levels of NCS. The meaning of the deviations between field experts’ apprehensions of cognitive structures and structures generated by the applied method should provide incentives for further research. It is clear, however, that the deviations were not only about more or less fine graded partitions, but also signaled a difference of how research concepts are associated. The variations between fields with regard to the estimated relevance of clusters should be commented. The considerably much stronger links of NCS arrived at in Case 2 (Md. NCS=0.31) were reflected by a higher relevance of clusters ( 2 percent misplaced articles) in comparison with Cases 1 and 3. The difference between Cases 1 and Case 3 with regard to the relevance of clusters (7 percent misplaced articles in Case 1 vs. 13 percent misplaced articles in Case 3) could only tentatively be assigned to the difference of the median NCS as this difference was not pronounced (0.15 in Case 1 and 0.09 in Case 3). The comparably less severe relative diminishing of the population in Case 156 2 specifically due to the clustering process (see Table 6-4) should preliminary be explained by the applied thresholds. 1.2 Case 4 The point of departure in the following discussion is in the final set of core documents containing 4,477 articles. This set was accomplished by a gradual reduction of the original set of 6,060 core documents by 26 percent when thresholds of NCS and cluster size were applied. 1.2.1 The Extent of Fragmentation Imposed by the Applied Method It was shown that the applied method leads to a fragmentation of research themes. On the average, a core document cluster could increase its size by a factor of eight and only a few clusters were expanded by less than half their sizes. The effect of fragmentation was illustrated by example where it also was shown that the adding of core documents brings about a decrease of cluster coherence, measured as D. Hence, the expansion of clusters is at a cost of a presumably diminished relevance. 1.2.2 The Impact of Iterated Clustering on the Overall Cluster Structure With the starting point in a large set of smaller clusters, the fusion of clusters at two subsequent levels showed an increasing loss of core documents as larger aggregations of core documents were formed. This loss was due to the generation of singleton clusters and isolated clusters emerging at each level of cluster fusion. At each subsequent level of cluster fusion, the general tendency was that by the increase of cluster size, there was a simultaneous decrease of cluster coherence and an increase of the external isolation of clusters. This means that increasingly less relevant clusters were formed but also that clusters got more isolated. 1.2.3 The Optimal Level of Cluster Fusion It was found that the second level of cluster fusion (C2) should be the optimal level. This could be the concluded on the following grounds: i. On the first level of cluster fusion (Cl), a large share of clusters were associated with other clusters through relatively strong links. ii. On the second level of cluster fusion, the internal coherence remained strong and at the same time clusters were generally more isolated. iii. On the last level of cluster fusion, the drop of cluster coherence was considerable, indicating the generation of more subject inconsistent clusters. Field experts’ evaluations did not contradict these findings. 157 1.2.4 Implication of Findings Though empirical findings speak in favor for the second level of cluster fusion as the most appropriate, it was shown that a few clusters on the Cl-level are nearly complete in terms of extrinsic associations and that a few clusters on the C3-level may be relevant. The example of cluster fusion over three levels illustrated that the association between disciplines through their research foci may provide interesting links which may give an overview of a problem area. The proposed method is, however, not likely to be applicable on the last level of cluster fusion (C3). Moreover, the quite severe loss of core documents generated by iterated clustering would require the interpretation of data from the preceding levels if a more comprehensive mapping should be accomplished. Hence, it is suggested that at least the two first levels of fusion are applied, including singleton clusters and isolated clusters and that mapping results be interpreted from bottom to top (or top bottom) as the cluster merging itself contain important information. The assessed effects of fragmentation implies that the proposed method when applied for core document clustering do not identify and map research themes exhaustively, but rather smaller cores of referencing consensus. Also, findings showed that approximately a quarter of the final population of core documents where lost when clustered, given the applied minimum size of clusters. 1.3 Reflections on Findings in Relation to Previous Research Several results connect to previous findings and theoretical considerations in the literature on bibliographic coupling and cocitation cluster analysis. First, claims that the method of bibliographic coupling is capable of associating documents that have a similar research focus (e.g. Vladutz & Cook, 1984; Peters, Braam & van Raan, 1995) was confirmed by the relatively high degree of relevance in clusters generated by the proposed method. The application of the complete link method, in line with coupling criterion B, originally suggested by Kessler ( 1960) and the suggestion of “cliques” as one type of bibliographically coupled document groups (Sen & Gan, 1983), resulted in small but compact and generally subject consistent clusters. Hence, the problems of “chaining” encountered in cocitation cluster analysis (cf. Griffith, Small, Stonehill & Dey, 1974) was avoided. The effect of fragmentation, or more precisely, the split up of research themes in smaller clusters, also encountered in cocitation cluster analysis (Braam, Moed & van Raan, 1991), was conspicuous. The issue of fragmentation is also related to the setting of thresholds of coupling strength. These problems was approached by Small and co-workers (Small & Sweeney, 1985) by implementing variable level clustering in order to find the best cluster solution. The problems of threshold setting was avoided in the case of core document mapping, as previous empirical findings would guide the setting of these (cf. Glänzel & Czerwon, 1995; 1996). The effect of the dependency of consensual referencing and the associated problems of threshold setting was observed in this study and could be 158 recognized as a severe diminishing of document populations. This type of problem has also been recognized in research in cocitation clustering where findings have shown that only parts of document populations relevant to identified research topics were revealed (cf. Braam, Moed & van Raan, 1991). This concerns the issue of the exhaustiveness of citation based science mapping. Braam, Moed and van Raan recognized that this question demands a comparison of cluster solutions on different levels of thresholds and the simultaneous use of complementary methods (ibid.). Through the criticism of the cocitation cluster analytical method, the statistical instability of the method (Oberski, 1988) and inconsistent results (Leydesdorff, 1987), the much varying results possible to arrive at by just tampering one of several affecting variables were highlighted. This connects to the difficulty to empirically arrive at method applications of citation based mapping that may generate optimal results, which could be illustrated as follows. Assume that the three more important variables are selected for empirical testing. Let these variables be the following ones: i. population; ii. choice of cluster method; and iii. threshold of coupling strength. Next, to each of these variables is assigned three sub-variables. The number of research settings required should then be 27. It would also be reasonable to include other multivariate techniques, e.g. factor analytical approaches, which should increase the number of research settings further. The difficulty to theoretically contribute to successful applications of citation based mapping is the absence of a conceptual framework. The significance of this problem was stressed by field experts’ comments concerning alternative (and equally valid) mapping solutions and by the fact that the proposed method generated generally relevant clusters much deviating from field experts’ clusters. Conclusively, a general problem of citation based science mapping is the absence of a common frame of reference. Therefore, more axiomatic approaches may pay off. 159 2. CONCLUSIONS In the study undertaken, a method was suggested for science mapping purposes and evaluated. The suggestion of this method was motivated by the fact that the prevailing method of citation based science document mapping, the cocitation cluster analytical method, can not map the most current published research, a feature that is a characteristic of the proposed method. The cocitation cluster analytical method, on the other hand, is based on a theory which claims that the more central research questions of a specialty can be identified through highly cocited documents. On this ground, it is presumed that the identification of the cognitive structures of specialties may be mapped. This is a feature that could not (as for now) be assigned to the proposed method. It was therefore assumed that none of these methods could substitute each other and that they would be complementary. Previous research has stated the capability of the bibliographic coupling method to associate subject similar documents with one another and its applicability for IR purposes. However, there is an explicit lack of empirical experience concerning the application of bibliographic coupling in the context of science mapping. Therefore, empirical experience from cluster analytical research in the context of science mapping could only be obtained from the research in cocitation cluster analysis. Based on criticism of the cocitation cluster analytical method and on reported empirical experiments, the following problems were presumed to be of importance also for the application of the proposed method: i. The dependency of consensual referencing implies that only minor shares of original document populations will be available for analysis. ii. The lack of a method for the decision of appropriate thresholds of coupling strength implies arbitrary threshold settings. iii. The choice of the single link cluster method has shown the undesirable effect of chaining (prolonged and loosely bound clusters) and the subsequent generation of macro clusters. iv. The partition of document populations has brought about the split up of research specialties, an effect of fragmentation of research fields. Findings confirmed the relevance of each of the above points with regard to the proposed method. These issues may be regarded as general for citation based science mapping applying documents as the analyzed unit. With regard to (i), only a minor fraction of the original populations were available for analysis and the stepwise diminishing of document populations was due to: (1) the filtering out of articles lacking bibliographic coupling relations; (2) the setting of thresholds of coupling strength and (3) the partition of document populations by the applied cluster method in combination with a set minimal cluster size. These three causes of reduction of populations all reflected the impact of and dependency on consensual referencing. 160 With regard to (ii), no valid method concerning the setting of thresholds of coupling strength was arrived at during the experimental phase. For the three first cases (single field level) considerations were taken with regard to the publication output of corresponding fields, and a ‘rule of thumb’ approach was applied. Findings showed that the application of severe thresholds of coupling strength implies more relevant clusters, but also that lower levels of coupling strength, necessary to apply in research settings where smaller or younger fields are mapped, are feasible. With regard to the fourth large and multidisciplinary research setting, the specific objective of mapping core documents implied the application of strict rules for the setting of thresholds. With regard to (iii), the design of the proposed method concerning the choice of cluster method was mainly based on theoretical considerations derived from the statistical literature on cluster analysis and the reported problems and criticism of the use of the single link cluster method in cocitation cluster analysis. Findings showed that the choice of the complete link cluster method resulted in coherent and generally subject consistent clusters. The well known drawback of the single link cluster method was hence steered clear of. However, as mentioned, the strict rules of merging also implied the generation of a large share of smaller sized clusters that from an information provision point of view should be regarded as noise, and the fdtering out of these added on to the aforementioned reduction of document populations. With regard to (iv), the effect of fragmentation was also seen in this study. For the first three cases, this effect was foremost noticed as a decisive difference of number of clusters and cluster sizes between partitions generated by field experts and partitions accomplished by the application of the proposed method. This difference was concluded and summarized applying Pratt's measure of concentration. Concerning the fourth case, the specific properties of core documents and the properties of the proposed method implied a research design aiming at the explicit elaboration of the easily foreseen effect of fragmentation. It was shown that core document clusters to a large extent depicted smaller consensual cores of current published research and that such cores could be expanded considerably also when only strong links were applied. This finding clearly showed that core document clusters constituted minor shares of larger research themes. It was further illustrated how iterated clustering could connect such cognitively related cores, and the optimal level of iterated clustering was found. The external points of reference accomplished by field experts’ evaluations of four examples of iterated clustering did not contradict these findings. Due to the complex relation between fragmentation and relevance (subject consistency) of core document clusters, it could be concluded that the information inherent on all levels of cluster fusion as well as in the process of cluster fusion itself should be used for optimal results. In the first three research settings relevant clusters were generally generated and it can be concluded that the proposed method has the capability to identify and map current and coherent research themes of a single research field, also when less severe thresholds of coupling strength are applied. The significance 161 of the information contained in the generated clusters was, however, not unambiguous as it was shown that the proposed method generated clusters strongly deviating from field expert’s conceptions of their own fields’ structures. This indicates that the proposed method generates information not anticipated by field experts and that this information may have a value of novelty. From another point of view, more congruence between field experts’ clusters and clusters generated by the proposed method may have indicated the possibility of replication of expert knowledge and opened up for new lines of research more connected to the elaboration of cognitive and social structures of science. In the fourth research setting, findings talks in favor for the generation of generally subject consistent clusters on the two first levels of cluster fusion. Findings also indicated that on the third level of cluster fusion, most significant links were exhausted, indicating the upper limit of cluster fusion for the proposed method. It could be assumed that sometimes useful information may be obtained also on this level. It could be concluded that the proposed method does not apply to traditional mapping objectives, i.e. the elucidation of specialty cognitive structures. Hence, its areas of application should foremost pertain to scientific information provision and be complementary to traditional citation indexing and cocitation cluster analysis. Further developments of the proposed method in the context of core document mapping and information provision could be accomplished. In particular, it could be suggested that the proposed method could be used as a navigating and information seeking tool. Several applications may be successful and one can be outlined on basis of findings. With a starting point in a complete graph (a core document cluster), the additional expansion by significant links could be used to monitor the radiating associations of articles related to a specific research theme. When additional information of cluster affiliation of such associated articles is added, the navigation in and between scientific structures would be facilitated.62 The navigation could be geared by varying threshold settings, deciding the maximum radius from each core. 62 This is actually the basic principle on which the database underlying the empirical study of Case 4 was built. The original idea of using core documents to trace links of associated articles and the subsequent mapping of research fronts was first presented by Glänzel & Czerwon (1995). Hence, the idea presented here is only a modification and expansion of their original idea. 162 REFERENCES Ahlgren, P. Jarneving, B. & Rousseau, R. (2003). Requirements for a co-citation similarity measure, with special reference to Pearson’s correlation coefficient. Journal of the American Society for Information Science & Technology. Aldenderfer, M.S. & Blashfield, R. K. (1984). Cluster analysis. In Quantitative Applications in the Social Sciences, Vol. 44. London: Sage Publications Inc. Baeza-Yates, R. & Ribeiro-Neto, B. (1999). Modern information retrieval. New-York: ACM Press. Biglan, A. (1973). The difference of subject matter in different academic areas. Journal of Applied Psychology. 57(3): 195-203. Braam, R. R., Moed, H. F. & van Raan, A. J. F. (1988). Mapping of science: Critical Elaboration and new approaches, a case study in agricultural biochemistry. In International conference on bibliometrics and theoretical aspects of information retrieval. (Eds. Egghe, L. and Rousseau, R.). Amsterdam: Elsevier Science Publishers. Braam, R. R., Moed, H. F. & van Raan, A. F. J. (1991). Mapping Science by combined Co-citation and word analysis 1 : structural aspects. Journal of the American Society for Information Science, 42(4):233-251. Cole, J.R. & Cole, S.C. (1973). Social Stratification in Science. Chicago: The University of Chicago Press. Egghe, L. & Rousseau, R. (1990). Introduction to informetrics. Amsterdam: Elsevier. Everitt, B.S., Landau, S. & Leese, M. (2001). Cluster analysis. Fourth edition. London: Arnold. Fano, R.M. (1956). Document in action. New York: Reinhold Publishing Corporation. Garfield, E. (1979). Citation Indexing: its theory and application in science, technology, and humanities New York: John Wiley & Sons. Garfield, E. (1998). From citation indexes to informetrics: Is the tail wagging the dog? Libri, 48, (2): 67-80. Glänzel, W. & Czerwon, H. J. (1995). A new methodological approach to bibliographic coupling and its application to research-front and other core documents, Proceedings of 5lh International Conference on scientometrics and Informetrics, held in River Forest, Illinois, June 7-10: 167-176. Glänzel, W. & Czerwon, H. J. (1996). A new methodological approach to bibliographic coupling and its application to the national, regional and institutional level. Scientometrics, 37(2): 195-221. 163 Griffith, B. Small, H. Stonehill, J. & Dey, S. (1974). The structure of scientific literatures II: toward a macro- and microstructure for science. Science studies, 4: 339-365. Jarneving, B. (2001). The cognitive structure of current cardiovascular research. Scientometric, 50 (3): 365-389. Johnsonbaugh, R. (1997). Discrete Mathematics. Prentice Hall International: New Jersey, Forth Edition. Johnson, D. E. (1998). Applied multivariate methods for data analysts. Pacific Grove: Duxbury Press. Kessler, M. M. (1958). Concerning some problems of intrascience communication. Massachusetts Institute for Technology, Lincoln Laboratory. Kessler, M. M. (1960). An experimental communication center for scientific and technical information. Massachusetts Institute for Technology, Lincoln Laboratory. Kessler, M. M. (1962). An experimental study of bibliographic coupling between technical papers. Massachusetts Institute for Technology, Lincoln Laboratory. Kessler, M.M. (1963a). Bibliographic coupling between scientific papers. American Documentation, 14(1): 10-25. Kessler, M.M. (1963b). Bibliographic coupling extended in time: Ten case histories. Information Storage and Retrieval, 1:169-187. Kessler, M.M. (1965). Comparison of the results of bibliographic coupling and analytic subject indexing. American Documentation, 16(3):223-233. Kruskal, J. B. (1964). Multidimensional scaling by optimizing goodness of fit to a nonmetric hypothesis. Psychometrica, 29 (1): 1-27 Kruskal, J. B., Wish, M. B. (1978). Multidimensional scaling. In Quantitative Applications in the Social Sciences, vol. 11. London: Sage Publications, Inc. Leydesdorff, L. (1987). Various methods for the mapping of science. Scientometrics. (ll):295-324. Marshakova, I.V. (1973). System of document connections based on references. Nauchno-Tekhnicheskaya Informatsiya Seriya 2-Informatsionnye Protsessy I Sistemy, 2(6):3-8. MacRoberts, M. H. & Mac Roberts B. R. (1989). Problems of citation analysis: a critical review. Journal of the American Society for Information Science. 40(5):342-349. Martyn, J. (1964). Bibliographic coupling. Journal of Documentation, 20:236. 164 McCain, K. W. (1986). Cocited author mapping as a valid representation of intellectual structure. Journal of the American Society for Information Science, 37(3):111-122. McCain, K..W. (1990). Mapping authors in intellectual space - a technical overview. Journal of the American Society for Information Science, 41(6):433-443. Miller, G.A. (1969). A psychological method to investigate verbal concepts. Journal of Mathematical Psychology. 6:169-191. Mubeen, M.A. (1995). Bibliographic coupling: an empirical study of economics. Annals of Library Science and Documentation, 42(2):41 -53. Noyons, E.C.M. (1999). Bibliometric mapping as a science policy and research management tool. Leiden: DSWO Press. Oberski, J.E.J. (1988). Some statistical aspects of co-citation cluster analysis and a judgement by physicists. In Handbook of quantitative studies of science and technology. Ed.:A.F.J.van Raan. Amsterdam: North Holland. Otte, E. & Rousseau, R. (2002). Social network analysis: a powerful strategy, also for the information sciences. Journal of Information Science. 28(6):441-453. Persson, O. (1988). Measuring Scientific output by online techniques. In Handbook of quantitative studies of science and technology. Ed.:A.F.J.van Raan. Amsterdam: North Holland. Persson, O. (1994). The intellectual base and research front of JASIS 1986-1990. Journal of the American Society for Information Science. 45( 1 ):31 -38. Peters, H. P. F., Braam, R. R. and van Raan, A. F. J. (1995). Cognitive resemblance and citation relations in chemical engineering publications. Journal of the American Society for Information Science. 46( 1 ):9-21. Pratt, A. D. (1977). A measure of class concentration in bibliometrics. Journal of the American Society for Information Science. 28, September: 285-292 Pritchard, Alan (1969). Statistical bibliography or bibliometrics? Journal of Documentation. 25(4):348-349. Sen, S. K. & Gan. S. K. (1983). A mathematical extension of the idea of bibliographic coupling and its applications. Annals of Library Science and Documentation, 30(2):78-82. Sharabchiev, Y.T. (1988). Comparative analysis of 2 methods of cluster analysis of bibliographic citation. (In Russian). Naucno-techniceskaja informacija I 2. Sharada, B.A. & Shanna, J.S. (1993). A study of bibliographic coupling in linguistic research. Annals of Library Science and Documentation, 40 (4). 25-137. 165 Small, H. (1973). Co-citation in the scientific literature: a new measure of the relationship between two documents. Journal of the American Society for Information Science, 24(July-August):265-269. Small, H. (1977). A co-citation model of a scientific specialty: A longitudinal study of collagen research. Social studies of science, 7(2): 139-166. Small, H. & Griffith, B. (1974). The structure of scientific literatures I: identifying and graphing specialities. Science studies, 4(1): 17-40. Small, H. & Griffith, B. (1983). The structure of the social and behavioral sciences literature. Stockholm papers in library and information science. Ed.: Stephan Schwarz. Stockholm: Royal Institute of Technology Library. TRITA-LIB- 6021. July. Small, H. Sweeney, E. (1985). Clustering the Science Citation Index using cocitations I. A comparison of methods. Scientometrics, 7(3-6):391 -409. Smith, L. (1981). Citation analysis. Library Trends, 30(1 ):83-105. Tijssen, R.J.W. (1992). Cartography of science', scientometric mapping with multidimensional methods. Leiden: DSWO Press, Leiden University. Vladutz, G. & Cook, J. (1984). Bibliographic coupling and subject relatedness. Proceedings of the ASIS Annual Meeting, AT. 204-207. Weinberg, B.H. (1974). Bibliographic coupling: a review. Information Storage and Retrieval. 10(5/&): 189-195. White, H. D. & Griffith, B. C. (1981). Author Cocitation: a literature measure of intellectual structure. Journal of the American Society for Information Science. 32:163-171. White, H. D. & McCain, K. W. (1998).Visualizing a Discipline: an author cocitation analysis of information science, 1972-1995. Journal of the American Society for Information Science. 49(4): 327-335. Ziman, John. (1984). An introduction to science studies: the philosophical and social aspects of science and technology. Cambridge: Cambridge University Press. 166 APPENDIX 1 EQUATIONS In this appendix all equations are gathered and the context in which they occur in the text of the study is briefly quoted when motivated. 2.1 The number of r-combinations of a set of n distinct elements is denoted by CO?, r) or 'n] , . P(n,r) n(n -1)• • ■ (n -r + 1) nl and ((/?./•) = = = rj r\ r\ (n-r)\r\ This equation is applied to assess the share of all possible pairs of objects (articles or clusters) in a defined set that are bibliographically coupled (the number of coupled pairs divided by C(n, r) where r = 2). It was for instance applied to measure the density of matrixes of bibliographic couplings of the final document populations. 2.2 This equation is applied when assessing the degree of interconnectedness, the density (£>), in sets of objects that may be described as graphs. D is defined as: D. :2-(#¿(G) A(tV-I) ’ where #L(G)= the number of edges connecting two vertices; and N= the number of vertices (Otte & Rousseau, 2002). The interval is [0, 1] and the maximum value is reached when the value of #L(G) equals the value of A(7V-l)/2. 63 This is similar to applying the next equation 2.2 though 2.2 is quoted and presented in a graphtheoretical context. 167 2.3 The Coupling Angle (C.A.) is expressed as: (^•^) ' D(ffD„k • Dok) C.A. is the coupling angle for citing documents j and k. DOj and Dok are the binary vectors of document j respectively k. The C.A. takes the maximum value of 1 if two Boolean vectors are parallel and 0 if they are rectangular. 2.4 The Jaccard coefficient (commonly referred to as the Jaccard’s index) is a well- known measure of the similarity .S' between two objects A and B, which counts the number of common attributes divided by the number of attributes possessed by at least one of the two objects: a n b In the context of cocitation analysis, this function is expressed as: 2.5 ACS/; (q+ç-c,,) 2.6 In the context of cocitation analysis, the cosine function (commonly referred to as Saltón’s cosine formula) is expressed as: ACS, where: NCSy = the normalized coupling strength between document i and/; Cij = the number of cocitations of document z and j; Cj = the number of citations of document z; and Cj = the number of citations of document j. All three equations, 2.4-2.6, take values in the interval [0,1], 168 4.1 The C.A. (2.3) was in practice calculated as: NCS = r¡J where NCSÿ = the normalized coupling strength between article z and article j ry = number of references common to both z and j ri, = number of references in the reference list of article i rij = number of references in the reference list of article j The interval is [0, 1 ] and n, = nj = r,7 gives the maximum value. This equation is referred to as the normalized coupling strength (NCS) in the text. 4.2 A measure of the internal cluster coherence is the Average Coupling Strength, AvgCS(C), for a cluster C. It is defined as: Z£cS(W AvgCS(C) =, n where n = number of articles in a cluster c, CS = number of bibliographic coupling units between two articles, d¡, d¡ and d,d^ C) This equation is complementary to equation 2.2 as these two measures of cluster coherence reflect different aspects of internal cluster coherence. 169 4.3 This equation is applied for the measuring of the distance (similarity) between two clusters. When calculating the average distance between clusters in a set of clusters (resulting from a partition) changes of cluster isolation can be monitored. Let C and C be clusters of sizes k and m, respectively. The average coupling strength between two clusters, C and C, AvgCSfC, C), is defined as: AvgCS(C,C') = kxm where CS = number of bibliographic coupling units between two articles, d¡, d¡ and di g C,<7/ g C 4.4 The concentration of articles to clusters was assessed applying Pratt 's measure of concentration. This measure is of general use when one wants to see how concentrated or spread out items (here articles) are when partitioned into categories (here clusters). Pratt’s measure is given as: c 2|((77 +1)/2)- 3 IN CASE 1 Bibliographic data of articles is presented in the following order: record number/ first author name/ publication year/ Journal name/ title/ author key words/key words plus. Missing data is indicated by “No Field”. CLUSTER 1 41/BURRELL QL/2002/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/ WILL THIS PAPER EVER BE C1TED/NO FIELD/KEYWORDS PLUS: LIBRARY CIRCULATION MODEL 117/BURRELL QL/2002/SCIENTOMETRICS /THE NTH-CITATION DISTRIBUTION AND OBSOLESCENCE/NO FIELD/KEYWORDS PLUS: LIBRARY CIRCULATION MODEL 169/BURRELL QL/2001/ SCIENTOMETRICS / STOCHASTIC MODELLING OF THE FIRST-CITATION DISTRIBUTION/NO FIELD/KEYWORDS PLUS: LIBRARY CIRCULATION MODEL; OBSOLESCENCE; GROWTH CLUSTER 3 36/CHEN CM/2002/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/V1SUALIZING AND TRACKING THE GROWTH OF COMPETING PARADIGMS - 2 CASE-STU DIES/NO FIELD/KEYWORDS PLUS: AUTHOR COCITATION; INTELLECTUAL STRUCTURE; CO- CITATION; SCIENCE; SPACES; VIBE 176/SMALL H/2001/ SCIENTOMETRICS/BELVER AND HENRY/NO FIELD/KEYWORDS PLUS: SCIENTIFIC LITERATURES; CO-CITATION; SCIENCE 196/KOEHLER W/2001/ SCIENTOMETRICS/INFORMATION-SCIENCE AS LITTLE SCIENCE - THE IMPLICATIONS OF A BIBLIOMETRIC ANALYSIS OF THE JOURNAL-OF-THE-AMERICAN- SOCIETY- FOR-INFORMATION-SC1ENCE/NO FIELD/KEYWORDS PLUS: SCIENTIFIC LITERATURE; CITAI ION ANALYSIS; AUTHORSHIP; LIBRARY; JAS1S; COCITATION; COUNTRIES 210/JARNEVING B/2001/ SCIENTOMETRICS/THE COGNITIVE STRUCTURE OF CURRENT CARDIOVASCULAR RESEARCH/NO FIELD/KEYWORDS PLUS: SCIENTIFIC LITERATURES; SCIENCE CLUSTER 4 19/CRONIN B/2001/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/HYPER AUTHORSHIP - A POST MODERN PERVERSION OR EVIDENCE OF A STRUCTURAL SHIFT IN SCHOLARLY COMMUNICATION PRACTICES/NO FIELD/KEYWORDS PLUS: MULTIPLE AUTHORSHIP; COLLABORATION; ARTICLES; SCIENCE; ORDER; ACKNOWLEDGMEN TS; BIBLIOMEI RICS; DISSEMINATION; CO-AUTHORSHIP; CONTRIBU TORS 45/CRONIN B/2001/ JOURNAL OF DOCUMENTATION/ACKNOWLEDGMENT TRENDS IN THE RESEARCH LITERATURE OF INFORMATION-SCIENCE/NO FIELD/KEYWORDS PLUS: COLLABORATION; SOCIOLOGY 106/CRONIN B/2002/ SCIENTOMETRICS/IDENTITY-CREATORS AND IMAGE-MAKERS - USING CITATION ANALYSIS AND THICK DESCRIPTION TO PUT AUTHORS IN THEIR PLACE/NO FIELD/KEYWORDS PLUS: SOCIOLOGY CLUSTER 14 72/GARG KC/2002/ SCIENTOMETRICS/SC1ENTOMETRICS OF LASER RESEARCH IN INDIA DURING 1970-1994/NO FIELD/KEYWORDS PLUS: SCIENCE; TECHNOLOGY; COLLABORATION; INDICATORS 80/GARG KC/2002/ SCIENTOMETRICS/SCIENTOMETRICS OF LASER RESEARCH IN INDIA AND CHINA/NO FIELD/KEYWORDS PLUS: CITATION PATTERNS; PUBLICATION; SCIENCE: IMPACT 171 110/HARITASH N/2002/ SCIENTOMETRICS/MAPPING OF S-AND-T ISSUES IN THE INDIAN PARLIAMENT - A SCIENTOMETRIC ANALYSIS OF QUESTIONS RAISED IN BOTH HOUSES OF THE PARLIAMENT/NO FIELD/KEYWORDS PLUS: INDICATORS; OUTPUT 189/GARG KC/2001/ SCIENTOMETRICS/A STUDY OF COLLABORATION IN LASER SCIENCE AND TECHNOLOGY/NO FIELD/KEYWORDS PLUS: INTERNATIONAL SCIENTIFIC COLLABORATION; POPULATION-GENETICS SPECIAL!! Y: SCIENTOMETRICS CLUSTER 16 76/LIANG LM/2002/ SCIENTOMETRICS/MAJOR FACTORS AFFECTING CHINA INTERREGIONAL RESEARCH COLLABORATION - REGIONAL SCIENTIFIC PRODUCTIVITY AND GEOGRAPHICAL PROXIMITY/NO FIELD/KEYWORDS PLUS: COOPERA HON 163/STEF ANIAK B/2001/ SCIENTOMETRICS/INTERNATIONAL-COOPERATION IN SCIENCE AND IN SOCIAL-SCIENCES AS REFLECTED IN MULTINATIONAL PAPERS INDEXED IN SCI AND SSCI/NO FIELD/KEYWORDS PLUS: COLLABORATION; COOPERATION; COUNTRIES; PROFILES 195/GLANZEL W/2001/ SCIENTOMETRICS/NAT1ONAL CHARACTERISTICS IN INTERNATIONAL SCIENTIFIC CO- AUTHORSI IIP RELATIONS/NO FIELD/KEYWORDS PLUS; COLLABORATION 219/GLANZEL W/2001/ SCIENTOMETRICS/DOUBLE EFFORT = DOUBLE IMPACT - A CRITICAL-VIEW AT INTERNATIONAL CO-AUTHORSHIP IN CHEMISTRY/NO FIELD/KEYWORDS PLUS: SCIENTIFIC COLLABORATION: SCIENCES; MODEL CLUSTER 18 4/LANGE LL/2002/ JOURNAL OF DOCUMENTATION/THE IMPACT FACTOR AS A PHANTOM - IS THERE A SELF-FULFILLING PROPHECY EFFECT OF IMPACT/AUTHOR KEYWORDS: VALUE ANALYSIS; ELECTRONIC PUBLISHING; DATABASES/KEYWORDS PLUS: JOURNAL IMPACT 131/V ANLEEUWEN TN/2002/ SCIENTOMETR1CS/DEVELOPMENT AND APPLICATION OF JOURNAL IMPACT MEASURES IN THE DUTCH SCIENCE SYSTEM/NO FIELD/KEYWORDS PLUS: CH ATIONS; INSTITUTE 126/GLANZEL W/2002/ SCIENTOME1RICS/JOURNAL IMPACT MEASURES IN BIBLIOMETRIC RESEARCH/NO FIELD/KEYWORDS PLUS: SCIENTIFIC LITERATURE; STOCHASTIC-MODEL; CITATION; INDICATORS; PRODUCTIVITY: INDEX CLUSTER 23 157/GURJEVA LG/2001/ SCIENTOMETRICS/SC1ENTOMETRICS IN THE CONTEXT OF PROBAB1LIS1IC PHILOSOPHY/NO FIELD/NO FIELD I60/NALIMOV VV/2001/ SCIENTOMETR1CS/CITATION-CLASS1CS OF NALIMOV,V.V - I - CURRENT­ CONTENTS. NUMBER 21. MAY 21, 1990/NO FIELD/NO FIELD 161/NALIMOV VV/2001/ SCIENTOMETRICS/CITATION-CLASSICS OF NALIMOV,V.V. 2 - CURRENT­ CONTENTS. NUMBER 24. JUNE 11. 1990 168/SHAPIRO SI/2001/SCIENTOMETRICS/THE UNIVERSE GRASPER/NO FIELD/NO FIELD CLUSTER 27 11/HUBER JC/2001/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/A NEW METHOD FOR ANALYZING SCIENTIFIC PRODUCTIV IT Y/NO FIELD/KEYWORDS PLUS: STATIONARY SCIENTOMETRIC DISTRIBUTIONS; CUMULATIVE ADVANTAGE; CREATIVITY; PARTICIPATION; PUBLICATION; STATISTICS; DURATION; SPEED 40/HUBER JC/2002/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/A NEW MODEL THAT GENERATES LOTKAS LAW/NO FIELD/KEYWORDS PLUS: SUCCESS-BREEDS-SUCCESS; INFORMETRIC DISTRIBUTIONS; CUMULATIVE ADVANTAGE; SCIENTIFIC PRODUCTIVITY; INVENTIVE PRODUCTIVITY: STATISTICS; RANDOMNESS: CREATIVITY; PUBLICATION: EXCEEDANCES 214/HUBER JC/2001/ SCIENTOMETRICS/SCIENTIFIC PRODUCTION - A STATISTICAL-ANALYSIS OF AUTHORS IN PHYSICS. 1800-I900/NO FIELD/NO FIELD 172 227/HUBER .IC/2001/ SCIENTOMETRICS/SCIENTIFIC PRODUCTION - A STATISTICAL-ANALYSIS OF AUTHORS IN MATHEMATICAL LOGIC/NO FIELD/KEYWORDS PLUS: STATIONARY SCIENTOMETRIC DISTRIBUTIONS; CUMULATIVE ADVANTAGE; PARTICIPATION; PUBLICATION; DURA TION: TESTS; SPEED; LAW CLUSTER 28 12/IVANCHEVA LE/2001/JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/THE NON-GAUSSIAN NATURE OF BIBLIOMETRIC AND SCIENTOMETRIC DISTRIBUTIONS - A NEW APPROACH TO INTERPRETA 1'ION/NO FIELD/KEYWORDS PLUS: PRODUCTIVITY 18/KRETSCHMER H/2001/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/AUTHOR INFLATION LEADS TO A BREAKDOWN OF LOTKAS LAW/NO FIELD/KEYWORDS PLUS: INFORMETRIC DISTRIBUTIONS; SCIENTIFIC COLLABORATION; PRODUCTIVITY; ATTRIBUTION; COUNTS 137/KARIS1DDAPPA CR/2002/ SCIENTOMETRICS/SCIENTIFIC PRODUCTIVITY OF AUTHORS IN THEORETICAL POPULATION- GENETICS/NO FIELD/KEYWORDS PLUS: FREQUENCY-DISTRIBUTION; LOTKAS LAW; PUBLICATION; TIME CLUSTER 36 94/LARSEN B/2002/ SCIENTOMETRICS/EXPLOITING CITATION OVERLAPS FOR INFORMATION­ RETRIEVAL - GENERATING A BOOMERANG EFFECT FROM THE NETWORK OF SCIENTIFIC PAPERS/NO FIELD/KEYWORDS PLUS: SYSTEMS; SCIENCE: DESIGN; WEB 183/SANDSTROM PE/2001/ SCIENTOMETRICS/SCHOLARLY COMMUNICATION AS A SOC1OECOLOGICAL SYSTEM/NO FIELD/KEYWORDS PLUS: HUMAN BEHAVIORAL ECOLOGY; CO­ CITATION; SCIENTIFIC LITERATURES; INTELLECTUAL STRUCTURE; INFORMATION- SEEKING; AUTHOR COCITATION; SCIENCE; RETRIEVAL; DOCUMENTS; SPACE 184/WHITE HD/2001/SCIENTOMETRICS/AUTHOR-CENTERED BIBLIOMETRICS THROUGH CAMEOS - CHARACTERIZATIONS AUTOMATICALLY MADE AND EDITED ONLINE/NO FIELD/KEYWORDS PLUS: CITATION ANALYSIS; PUBLICATIONS; RETRIEVAL; MODEL CLUSTER 38 10/LEYDESDORFF L/2001/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/THE SELF-ORGANIZATION OF THE EUROPEAN INFORMATION-SOCIETY - THE CASE OF BIOTECHNOLOGY/NO FIELD/KEYWORDS PLUS: CO-CITATIONS; SCIENCE; INDICATORS; GOVERNMENT; TECHNOLOGY: INDUSTRY; WORDS 30/LEYDESDORFF L/2002/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/DYNAMIC AND EVOLUTIONARY UPDATES OF CLASSIFICATORY SCHEMES IN SCIENTIFIC JOURNAL STRUCTURES/NO FIELD/KEYWORDS PLUS: BRITISH SCIENCE; BIBLIOMETRIC ASSESSMENT; DECLINE; PERFORMANCE; INDICATORS: NATIONS 141/LEYDESDORFF L/2002/ SCIENTOMETRICS/INDICATORS OF STRUCTURAL-CHANGE IN THE DYNAMICS OF SCIENCE - ENTROPY STATISTICS OF THE SCI JOURNAL-CITATION-REPORTS/NO FIELD/KEYWORDS PLUS: COMMUNICATION; INTELLIGENCE; PERFORMANCE; TECHNOLOGY; KNOWLEDGE; IMPACT; AREAS 220/VILANOVA MR/2001/ SCIENTOMETRICS/WHY CATALONIA CANNOT BE CONSIDERED AS A REGIONAL INNOVATION SYSTEM/NO FIELD/KEYWORDS PLUS: INDUSTRY-GOVERNMENT RELATIONS; PATENT STATISTICS; EUROPEAN-UNION; TRIPLE-HELIX; SCIENCE; TECHNOLOGY CLUSTER 41 89/VERBEEK A/2002/ SCIENTOMETRICS/LINKING SCIENCE TO TECHNOLOGY - USING BIBLIOGRAPHIC REFERENCES IN PATENTS TO BUILD LINKAGE SCHEMES/NO FIELD/KEYWORDS PLUS: INDICATORS 96/MEYER M/2002/ SCIENTOMETRICS/TRACING KNOWLEDGE FLOWS IN INNOVATION SYSTEMS/NO FIELD/KEYWORDS PLUS: PATENT CITATIONS; SCIENCE; TECHNOLOGY; INDICATORS; INVENTIONS; INDUSTRY; LINKAGE: US 173 199/MEYER MS/2001/ SCIENTOMETRICS/PATENT CITATION ANALYSIS IN A NOVEL FIELD OF TECHNOLOGY - AN EXPLORATION OF NANO-SCIENCE AND NANO-TECHNOLOGY/NO FIELD/KEYWORDS PLUS: TECHNICAL CHANGE CLUSTER 45 101/PERITZ BC/2002/ SCIENTOMETRICS/THE SOURCES USED BY BIBLIOMETRICS- SC1ENTOMETRICS AS REFLECTED IN REFERENCES/NO FIELD/KEYWORDS PLUS: SCIENTIFIC JOURNALS; CITATION ANALYSIS; SELF- CITATION; SCIENCE; PATTERNS; DECADES 133/SCHUBERT A/2002/ SCIENTOMETRICS/THE WEB OF SCIENTOMETRICS - A STATISTICAL OVERVIEW OF THE 1ST 50 VOLUMES OF THE JOURNAL/NO FIELD/NO FIELD 225/SCHOEPFLIN U/2001Z SCIENTOMETRICS/2 DECADES OF SCIENTOMETRICS - AN INTERDISCIPLINARY FIELD REPRESENTED BY ITS LEADING JOURNAL/NO FIELD/KEYWORDS PLUS: SCIENCES CLUSTER 47. 23/WHITE HD/2001/ JOURNAL. OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/AUTHORS AS CITERS OVER TIME/NO FIELD/KEYWORDS PLUS: CITAIION ANALYSIS; ORTEGA HYPOTHESIS; INFORMATION-SCIENCE; SELF-CITATIONS; MOTIVATIONS; KNOWLEDGE; MODEL; CLASSIFICATION; DOCUMENTA TION; REFERENCES 25/WHITLEY KM/2002/ JOURNAL OF 'HIE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/ANALYSIS OF SCIFINDER SCHOLAR AND WEB OF SCIENCE CITATION SEARCHES/NO FIELD/NO FIELD 102/PICHAPPAN P/2002/ SCIENTOMETRICS/THE OTHER SIDE OF THE COIN - THE INTRICACIES OF AUTHOR SELF- CITATIONS/NO FIELD/KEYWORDS PLUS: SCIENCE; COMMUNICATION; BEHAVIOR; LEVEL CLUSTER 48 103/PRIME C/2002/ SCIENTOMETRICS/COCITATIONS AND CO-SITATIONS - A CAUTIONARY VIEW ON AN ANALOGY/NO FIELD/KEYWORDS PLUS: SCIENTIFIC LITERATURE; SCIENCE; COCITATION; IMPACT 206/SCHWECHHEIMER H/2001/ SCIENTOMETRICS/MAPPING INTERDISCIPLINARY RESEARCH FRONTS IN NEUROSCIENCE - A BIBLIOMETRIC VIEW TO RETROGRADE-AMNESIA/NO FIELD/KEYWORDS PLUS: CO-CITATIONS; SCIENCE 224/SALZARULO L/2001/ SCIENTOMETR1CS/BIAS, STRUCTURE AND QUALITY IN CITATION INDEXING/NO FIELD/KEYWORDS PLUS: CO-CITATIONS; SCIENCE CLUSTER 51 2/THELWALL M/2002/JOURNAL OF DOCUMENTATION/EVIDENCE FOR THE EXISTENCE OF GEOGRAPHIC TRENDS IN UNIVERSITY WEB SITE INTERLINKING/AUTIIOR KEYWORDS: INTERNET; KNOWLEDGE WORKERS; UNIVERSITIES; UNITED K1NGDOM/KEYWORDS PLUS: CITATION ANALYSIS; IMPACT FACTORS; INFORMA TION; INTERNET; SCIENCE; CRAWLER 5/THELWALL M/2002/ JOURNAL OF DOCUMENTATION /A COMPARISON OF SOURCES OF LINKS FOR ACADEMIC WEB IMPACT FACTOR CALCULATIONS/AUTHOR KEYWORDS: INTERNET; INFORMATION RETRIEVAL/KEYWORDS PLUS: CITATION; INFORMATION 8/THELWALL M/2001/JOURNAL OF INFORMATION SCIENCE/EXPLORING THE LINK STRUCTURE OF THE WEB WITH NETWORK DIAGRAMS/NO FIELD/KEYWORDS PLUS: IMPACT FACTORS; SEARCH 15/TTIELWALL M/2001/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/EXTRACTING MACROSCOPIC INFORMATION FROM WEB LINKS/NO FIELD/KEYWORDS PLUS: RESEARCH ASSESSMENT EXERCISE; WORLD-W1DE-WEB; IMPACT FACTORS; SCHOLARLY COMMUNICATION; UNIVERSITY DEPARTMENTS; CITATION COUNTS; SEARCH ENGINE; BRI TISH: CONTINUUM; ANATOMY 174 31/THELWALL M/2002/ JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY/CONCEPTUAL1ZING DOCUMENTATION ON THE WEB - AN EVALUATION OF DIFFERENT HEURISTIC-BASED MODELS FOR COUNTING LINKS BETWEEN UNIVERSITY WEB SITES/NO FIELD/KEYWORDS PLUS: IMPACT FACTORS: CITATION ANALYSIS: SEARCH ENGINE; COMMUNICATION; INTERNET; INFORMATION; CRAWLER; DESIGN; PAGES 46/THELWALL M/2001/ JOURNAL OF INFORMATION SCIENCE /A WEB CRAWLER DESIGN FOR DATA MINING/NO FIELD/KEYWORDS PLUS: IMPACT FACTORS; SEARCH ENGINE; SITE 62/THELWALL M/2002/ SCIENTOMETRICS/INTERLINKING BETWEEN ASIA-PACIFIC UNIVERSITY WEB SITES/NO FIELD/NO FIELD 87/SMITH A/2002/ SCIENTOMETR1CS/WEB IMPACT FACTORS FOR AUSTRALASIAN UNIVERS1TIES/NO FIELD/KEYWORDS PLUS: CO-AU WORSHIP CLUSTER 62 144/BEAVER DD/2001/ SCIENTOMETRICS/REFLECT1ONS ON SCIENTIFIC COLLABORATION. (AND ITS STUDY) - PAST, PRESENT, AND FUTURE/NO FIELD/KEYWORDS PLUS: CO-AUTHORSHIP 63/MARTINSEMPERE MJ/2002/ SCIENTOMETRICS /THE EFFECT OF TEAM CONSOLIDATION ON RESEARCH COLLABORATION AND PERFORMANCE OF SCIENTISTS - CASE-STUDY OF SPANISH UNIVERSITY RESEARCHERS IN GEOLOGY/NO FIELD/NO FIELD 69/FARAHAT H/2002/ SCIENTOMETRICS /AUTHORSHIP PATTERNS IN AGRICULTURAL SCIENCES IN EGYPT/NO FIELD/KEYWORDS PLUS: SCIENTIFIC CO-AUTHORSHIP; RESEARCH COLLABORATION; MULTIPLE AUTHORSHIP; LIBRARY 153/WAGNERDOBLER R/2001/ SCIENTOMETRICS /CONTINUITY AND DISCONTINUITY OF COLLABORATION BEHAVIOR SINCE 1800 - FROM A BIBLIOMETRIC POINT-OF-VIEW/NO FIELD/KEYWORDS PLUS: SCIENTIFIC CO-AUTHORSHIP 175 APPENDIX 3 THE COMPARISON OF TWO PARTITIONS IN CASE 1 The Dispersion of Articles over Clusters for Two Partitions. In the following table, columns A-D show the dispersion of articles in clusters generated by the field expert over the clusters generated by the complete link cluster method whereas columns E-H show the dispersion of articles in clusters generated by the complete link cluster method over the clusters generated by the field expert. A B C D E F G H Complete Complete Expert Expert Complete Complete Expert Expert doc.nr. clu.nr. doc.nr. clu.nr. doc.nr. clu.nr. doc.nr. clu.nr. 4 18 4 1 41 1 41 2 126 18 126 1 117 1 117 2 131 18 131 1 169 1 169 2 30 38 30 1 36 3 36 3 141 38 141 1 210 3 210 3 25 47 25 1 196 3 196 8 224 48 224 1 176 3 176 9 41 1 41 2 19 4 19 4 117 1 117 2 106 4 106 7 169 1 169 2 45 4 45 8 11 27 11 2 110 14 110 6 40 27 40 2 72 14 72 8 214 27 214 2 80 14 80 8 227 27 227 2 189 14 189 8 12 28 12 2 76 . 16 76 4 18 28 18 2 163 16 163 4 36 3 36 3 195 16 195 4 210 3 210 3 219 16 219 4 183 36 183 3 4 18 4 1 184 36 184 3 126 18 126 1 206 48 206 3 131 18 131 1 19 4 19 4 157 23 157 9 76 16 76 4 160 23 160 9 163 16 163 4 161 23 161 9 195 16 195 4 168 23 168 9 219 16 219 4 11 27 11 2 63 62 63 4 40 27 40 2 69 62 69 4 214 27 214 2 144 62 144 4 227 27 227 2 153 62 153 4 12 28 12 2 133 45 133 5 18 28 18 2 103 48 103 5 137 28 137 8 2 51 2 5 94 36 94 10 5 51 5 5 183 36 183 3 8 51 8 5 184 36 184 3 15 51 15 5 30 38 30 1 31 51 31 5 141 38 141 1 46 51 46 5 10 38 10 6 176 62 51 62 5 220 38 220 6 87 51 87 5 89 41 89 6 110 14 110 6 96 41 96 6 10 38 10 6 199 41 199 6 220 38 220 6 133 45 133 5 89 41 89 6 101 45 101 8 96 41 96 6 225 45 225 8 199 41 199 6 25 47 25 1 106 4 106 7 23 47 23 7 23 47 23 7 102 47 102 7 102 47 102 7 224 48 224 1 196 3 196 8 206 48 206 3 45 4 45 8 103 48 103 5 72 14 72 8 2 51 2 5 80 14 80 8 5 51 5 5 189 14 189 8 8 51 8 5 137 28 137 8 15 51 15 5 101 45 101 8 31 51 31 5 225 45 225 8 46 51 46 5 176 3 176 9 62 51 62 5 157 23 157 9 87 51 87 5 160 23 160 9 63 62 63 4 161 23 161 9 69 62 69 4 168 23 168 9 144 62 144 4 94 36 94 10 153 62 153 4 \n APPENDIX 4: BIBLIOGRAPHIC DESCRIPTIONS OF CLUSTERS WITH A SIZE > 3 IN CASE 2 Bibliographic data as follows: record number/ first author name/ publication year/ Journal name/ title/ author key words/key words plus. Missing data is indicated by “No Field”. CLUSTER 1 1014/SHI M/2002/ ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/CATALYTIC, ASYMMETRIC BAYLIS-HILLMAN REACTION OF IMINES WITH METHYL VINYL KETONE AND METHYL ACRYLATE/NO FIELD/KEYWORDS PLUS: TITANIUM(IV) CHLORIDE; LEWIS BASE; ALDEHYDES; PHOSPHINE; ESTERS 7888/SHI M/2003/ JOURNAL OF ORGANIC CHEMISTRY/AN UNEXPECTED HIGHLY STEREOSELECTIVE DOUBLE AZA-BAYLIS- HILLMAN REACTION OF SULFONATED IMINES WITH PHENYL VINYL KETONE/NO FIELD/KEYWORDS PLUS: LEWIS BASE; TITANIUM(IV) CHLORIDE: ALDEHYDES; PHOSPHINE 12254/SHI M/2002/ TETRAHEDRON LETTERS/ONE-POT AZA-BAYLIS-HILLMAN REACTIONS OF ARYLALDEHYDES AND DIPHENYLPHOSPHINAMIDE WITH METHYL VINYL KETONE IN THE PRESENCE OF TICL4PPH3, AND ET3N/NO FIELD/KEYWORDS PLUS: ELECTRON-DEFICIENT ALKENES; T1 TANIUM(IV) CHLORIDE; LEWIS BASE; ALDEHYDES; PHOSPHINE; ESTERS 13307/SHI M/2002/ TETRAHEDRON LETTERS/BAYL1S-HILLMAN REACTIONS OF N- ARYLIDENEDIPHENYLPHOSPHINAMIDES WITH METHYL VINYL KETONE. METHYL ACRYLATE. AND ACRYLONITRILE/AUTHOR KEYWORDS: N-ARYLIDENEDIPHENYLPHOSPHINAMIDE; LEWIS BASE; BAYLIS-HILLMAN REACTION; METHYL VINYL KETONE (MVK); METHYL ACRYLATE; ACRYLONITRILE/KEYWORDS PLUS: ELECTRON-DEFICIENT ALKENES; TJTANIUM(IV) CHLORIDE; LEWIS BASE; ALDEHYDES; PHOSPHINE; ESTERS 14376/SH1 M/2002/ TETRAHEDRON LETTERS/LEWIS BASE AND L-PROLINE CO-CATALYZED BAYLIS-HILLMAN REACTION OF ARYLALDEHYDES WITH METHYL VINYL KETONE/AUTHOR KEYWORDS; BAYLIS-HILLMAN REACTION; LEXIS BASE; METHYL VINYL KETONE (MVK); L- PROLINE; IMIDAZOLE; TRIETHYLAMINE/KEYWORDS PLUS: TITANIUM(IV) CHLORIDE; CONJUGATE ADDITION; ALDOL REACT IONS; ALDEHYDES CLUSTER 2 7415/C ASTRO EA/2003/ JOURNAL OF ORGANIC CHEMISTRY/KINETIC INVESTIGATION OF THE REACTIONS OF S-4-NITROPHENYL 4- SUBSTITUTED THIOBENZOATES WITH SECONDARY ALICYCLIC AMINES IN AQUEOUS-ETHANOL/NO FIELD/KEYWORDS PLUS: NUCLEOPHILIC- SUBSTITUTION REACT IONS; S-ARYL THIOCARBONATES; 4-NITROPHENYL THIONOCARBONATES; METHYL CARBONATE; ESTER AMINOLYSIS; MECHANISM; ACETONITRILE; 2.4- DINITROPHENYL; PHENYL; PYRIDINOLYSIS 7732/CASTRO EA/2003/ JOURNAL OF ORGANIC CHEMISTRY/KINETICS AND MECHANISM OF THE AMINOLYSIS OF 4-METHYLPHENYL AND 4-CHLOROPHENYL 2.4-DINITROPHENYL CARBONATES IN AQUEOUS- ETHANOL/NO FIELD/KEYWORDS PLUS: STRUCTURE-REACTIVITY CORRELATIONS: 2.4.6- TRINITROPHENYL METHYL CARBONATE; SECONDARY ALICYCLIC AMINES; S-ARYL THIOLCARBONATES; CONCERTED MECHANISM; SUBSTITUTED PYRIDINES; TRANSITION-STATE; ESTER AMINOLYSIS; ETHYL; ACETATE 8686/CASTRO EA/2002/ JOURNAL OF ORGANIC CHEMISTRY/KINETICS AND MECHANISM OF THE AMINOLYSIS OF METHYL 4- NITROPHENYL. METHYL 2.4-DINITROPHENYL, AND PHENYL 2.4- DINITROPHENYL CARBONATES/NO FIELD/KEYWORDS PLUS: SECONDARY ALICYCLIC AMINES; STRUCTURE- REACTIVITY CORRELATIONS; RATE-DETERMINING STEP; 2.4.6- TRINITROPHENYL ACETATE: SUBSTITUTED PYRIDINES; CONCERTED MECHANISMS: AQUEOUS-SOLUTION; PYRIDINOLYSIS; I l IIONOCARBONA1 ES: TI IIOCARBONATE 178 CLUSTER 3 7672/CASTRO EA/2003/JOURNAL OF ORGANIC CHEMISTRY/KINETIC-STUDY OF THE PHENOLYSIS OF O-METHYL AND O-PFIENYL O- 2,4-DINITROPHENYL THIOCARBONATES AND O-ETHYL 2,4- DINITROPHENYL DITHIOCARBONATE/NO FIELD/KEYWORDS PLUS: STRUCTURE-REACTIVITY CORRELATIONS; PHENOLATE ION NUCLEOPHILES; SUBSTITUTED PHENOXIDE IONS; SECONDARY ALICYCLIC AMINES; TRANSITION-STATE STRUCTURE; ACYL-TRANSFER REACTIONS; ACETYL GROUP TRANSFER; CONCERTED MECHANISMS; 4- NITROPHENYL CHLOROTHIONOFORMATES; OXYGEN NUCLEOPHILES 8098/CASTRO EA/2003/JOURNAL OF ORGANIC CHEMISTRY/KINETICS AND MECHANISM OF THE BENZENETHIOLYSIS OF 2,4- DINITROPHENYL AND 2.4.6-TRINH ROPHENYL METHYL. CARBONATES AND S-(2.4-DINITROPHENYL) AND S-(2,4,6-TRINITROPHENYL) ETHYL THIOLCARBONATES/NO FIELD/KEYWORDS PLUS: ACYL GROUP TRANSFER: STRUCTURE­ REACTIVITY CORRELATIONS; SUBSTITUTED PHENOXIDE IONS; S-ARYL THIOCARBONATES; CONCERTED MECHANISMS; TRANSITION-STATE; AQUEOUS-SOLUTION; OXYGEN NUCLEOPHILES: MECN-H20 MIXTURES; ESTER AMINOLYSIS 9348/CASTRO EA/2002/JOURNAL OF ORGANIC CHEMISTRY/KINETICS AND MECHANISM OF THE PHENOLYSIS OF ASYMMETRIC DIARYL CARBONATES/NO FIELD/KEYWORDS PLUS: STRUCTURE­ REACTIVITY CORRELATIONS; PHENOLATE ION NUCLEOPHILES; SECONDARY ALICYCLIC AMINES; SUBSTITUTED PHENOXIDE IONS; TRANSITION-STATE STRUCTURE; ACYL-TRANSFER REACTIONS; ACETYL GROUP TRANSFER: CONCERTED MECHANISMS; AQUEOUS-SOLUTION; AMINOLYSIS CLUSTER 4 10297/LHOTAK P/2003/TETRAHEDRON LETTERS/SYNTHESIS OF A DEEP-CAVITY THIACALIX(4)ARENE/AUTHOR KEYWORDS: THIACALIXARENES; X-RAY CRYSTALLOGRAPHY; ALKYLATION; CONFORMATIONAL ANALYSIS/KEYWORDS PLUS: STATE STRUCTURAL-ANALYSIS; SOLID-STATE; INFINITE CHANNELS; CALIX(4)ARENES; DERIVATIVES; RIM; P-TERT- BUTYLTHIACALIX(4)ARENE; THIACALIXARENE; CALIXARENES; CONFORMERS 10480/LHOTAK P/2003/TETRAHEDRON LETTERS/STEREOSELECTIVE OXIDATION OF THIACALIX(4)ARENES WITH THE NANO3/CF3COOH SYSTEM/NO FIELD/KEYWORDS PLUS: SELECTIVE OXIDATION; METAL-IONS: UPPER RIM; P- TERT-BUTYLT1 IIACALIX(4)ARENE; SULF1NYLCALIX(4)ARENES; CALIX(4)ARENES; DERIVATIVES; SULFINYL 12144/LHOTAK P/2002/TETRAHEDRON LETTERS/ALKYLATION OF THIACALIX(4)ARENES/NO FIELD/KEYWORDS PLUS: SOLID-STATE; INFINITE CHANNELS: P-TERT- BUTYLTHIACALIX(4)ARENE; CONFORMERS 12650/LHOTAK P/2002/TETRAHEDRON LETTERS/NITRATION OF THIACALIX(4)ARENE DERIVATIVES/NO FIELD/KEYWORDS PLUS: SELECTIVE OXIDATION; METAL-IONS; P-TERT- BUTYLTH1ACALIX(4)ARENE; SULFINYLCAL1X(4»ARENES; CALIXARENES 13490/LHOTAK P/2002/TETRAHEDRON LETTERS/DIAZO COUPLING - AN ALTERNATIVE METHOD FOR THE UPPER RIM AMINATION OF THIACALIX(4)ARENES/NO FIELD/KEYWORDS PLUS: P-TERT- BUTYLTHIACALIX(4)ARENE; CALIXARENES; OXIDATION CLUSTER 5 6342/LEWIS FD/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/DYNAMICS OF INTERSTRAND AND INTRASTRAND HOLE TRANSPORT IN DNA HAIRPINS/NO FIELD/KEYWORDS PLUS: CHARGE-TRANSPORT; DISTANCE; MECHANISM; OXIDATION 11268/TAKADA T/2003/TETRAHEDRON LETTERS/HOLE TRANSFER IN DNA - DNA AS A SCAFFOLD FOR HOLE TRANSFER BETWEEN 2 ORGANIC-MOLECULES/NO FIELD/KEYWORDS PLUS: DISTANCE CHARGE-TRANSPORT; TRANSIENT ABSORPTION; HOPPING MECHANISM; ELECTRON-TRANSFER; REDOX CHEMISTRY; RADICAL-CATION: DUPLEX DNA; OXIDATION; DYNAMICS; 2-AMINOPURINE 12470/KAWAI K/2002/TETRAHEDRON LETTERS/REGULATION OF ONE-ELECTRON OXIDATION RATE OF GUANINE AND HOLE TRANSFER RATE IN DNA THROUGH HYDROGEN- BONDING/AUTHOR KEYWORDS: DNA; HYDROGEN BONDING: ONE-ELECTRON OXIDATION; HOLE TRANSFER/KEYWORDS PLUS: DIIMIDE DERIVATIVES: TRANSPORT; DISTANCE; SEQUENCE; NAPH1 HALENE; CHEMISTRY; CLEAVAGE; IMIDE; GG 179 CLUSTER 6 723/GARTNER ZJ/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDIT1ON/2 ENABLING ARCHITECTURES FOR DNA-TEMPLATED ORGANIC-SYNTHESIS/AUTHOR KEYWORDS: BIOORGANIC CHEMISTRY: COMBINATORIAL CHEMISTRY; DNA: TEMPLATE SYNTHESIS/KEYWORDS PLUS: NUCLEIC-ACIDS; LIGATION: OLIGONUCLEOTIDES; AMPLIFICATION: REPLICATION: SYSTEM: ORIGIN; PNA 1112/CALDERONE CT/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/DIRECTING OTHERWISE INCOMPATIBLE REACTIONS IN A SINGLE SOLUTION BY USING DNA-TEMPLATED ORGANIC-SYNTHESIS/AUTHOR KEYWORDS: COMBINATORIAL CHEMISTRY; DIVERSIFICATION: OLIGONUCLEOTIDES; SYNTHETIC METHODS: TEMPLATE SYNTHESIS/KEYWORDS PLUS: NATURAL PRODUCT: LIGATION: AMPLIFICATION; MOLECULES 1579/GARTNER ZJ/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/EXPANDING THE REACTION SCOPE OF DNA-TEMPLA FED SYNTHESIS/AUTHOR KEYWORDS: COUPLING REACTIONS: DNA: MOLECULAR EVOLUTION; SYNTHETIC METHODS; TEMPLATE SYNTHESIS/KEYWORDS PLUS: REPLICA! ION; LIGATION: ACIDS; RNA CLUSTER 7 495/PIDATII ALA C/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/DIRECT CATALYTIC ASYMMETRIC ENOLEXO ALDOLIZATIONS/AUTHOR KEYWORDS: ALDOL REACTION: AMINO ACIDS; ASYMMETRIC CATALYSIS: ORGANOCATALYSIS/KEYWORDS PLUS: 3-COMPONENT MANNICH REACTION: DYNAMIC KINETIC RESOLUTION: ALDOL REACTIONS; AMINO-AC1DS; CARBOXYLIC ESTERS; ALPHA-AMINATION; PROLINE: ALDEHYDES; INDUCTION; HYDROGENATION 4105/BAH MANYAR S/2003/.IOURNAL OF THE AMERICAN CHEMICAL SOCIETY/QUANTUM- MECHANICAL PREDICTIONS OF THE STEREOSELECTIVITIES OF PROLINE-CATALYZED ASYMME 4'RIC INTERMOLECULAR ALDOL REACTIONS/NO FIELD/KEYWORDS PLUS: MOLECULAR- ORBITAL METHODS: GAUSSIAN-TYPE BASIS; ORGANIC-MOLECULES 4502/HOANG L/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/KINE TIC AND STEREOCHEMICAL EVIDENCE FOR THE INVOLVEMENT’ OF ONLY ONE PROLINE MOLECULE IN THE TRANSITION-STATES OF PROLINE-CATALYZED INTRAMOLECULAR AND INTERMOLECULAR ALDOL REACTIONS/NO FIELD/KEYWORDS PLUS: ASYMMETRIC-SYNTHESIS; MECHANISM; CYCLIZATION; KETONES CLUSTERS 12402/WALSH LM/2002/TETRAHEDRON LETTERS/SULFIDE-BF3-CENTER-DOT-OET2 MEDIATED BAYLIS-HILLMAN REACTIONS/NO FIELD/KEYWORDS PLUS: BETA-HYDROXY-KETONES; TITANIUM!IV) CHLORIDE; ALPI IA,BETA-UNSATURATED KETONES; ALDEHYDES; CATALYSTS 12575/CATR1 R/2002/TETRAHEDRON LETTERS/1MIDAZOLE-CATALYZED BAYLIS-HILLMAN REACTIONS - A NEW ROUIE TO ALLYLIC ALCOHOLS FROM ALDEHYDES AND CYCLIC ENONES/NO FIELD/NO FIELD 12730/KATAOKA T/2002/TETRAHEDRON LETTERS/!'AN DEM MICHAEL-ALDOL REACTION VIA 6- ENDO-DIG CYCLIZATION OF YNONE-CHALCOGENIDES - SYNTHESIS OF 2-UNSUBSTITUTED 3- (HYDROXYALKYDCHALCOGENOCHROMEN-4-ONES/NO FIELD/KEYWORDS PLUS: BAYLIS- HILLMAN REACTION; ELECTRON-DEFICIENT ALKENES; ALPHA.BETA-ACETYLENIC KETONES; ETHYNYL KETONES; ALDEHYDES; FLAVONES CLUSTER 9 534/STORER RI/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/A TOTAL-SYNTHESIS OF EPOTHILONES USING SOLID-SUPPORTED REAGENTS AND SCAVENGERS/AUTHOR KEYWORDS: ALDOL REACTION: ANTITUMOR AGENTS; NATURAL PRODUCTS: POLYKETIDES; POLYMERS/KEYWORDS PLUS: ASYMMETRIC ALDOL REACTION; ENANTIOSELECTIVE TOTAL SYNTHESIS: STEREOSELECTIVE TOTAL SYNTHESIS; MICROTUBULE-STABILIZING AGENTS; OLEFIN METATHESIS APPROACH; FORMAL TOTAL SYNTHESIS; CHIRAL LEWIS-ACID; ALKYNE METATHESIS; ORGANIC-SYNTHESIS; SOLUTION-PHASE 180 1631/SUN J/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/STEREOSELECTIVE TOTAL SYNTHESIS OE EPOTHILONES BY THE METATHESIS APPROACH INVOLVING C9-C10 BOND FORMATION/NO FIELD/KEYWORDS PLUS: MICROTUBULE-STABILIZING AGENTS; OLEFIN METATHESIS; B ANALOGS; TAXOL; CELLS 1640/LIU JJ/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/ALDOLASE-CATALYZED ASYMMETRIC-SYNTHESIS OF NOVEL PYRANOSE SYNTHONS AS A NEW ENTRY TO HETEROCYCLES AND EPO'l HILONES/AUTHOR KEYWORDS: ALDOL REACTION; ENZYME CATALYSIS; EPOTHILONES; SYNTHETIC METHODS; TOTAL SYNTHESIS/KEYWORDS PLUS: ENANTIOSELECTIVE TOTAL SYNTHESIS; 2- DEOXYRIBOSE-5-PHOSPHATE ALDOLASE; STEREOSELECTIVE SYNTHESIS; CHEMOENZYMATIC SYNTHESIS; (-)-EPOTHILONE-A; CONFORMATION; KETONES 8856/CHAPPELL MD/2002/JOURNAL OF ORGANIC CHEMISTRY/PROBING THE SAR OF DEPOB VIA CHEMICAL SYNTHESIS - A TOTAL SYNTHESIS EVALUATION OF C26-(1.3-DIOXOLANYL)-12,13- DESOXYEPOTHILONE-B/NO FIELD/KEYWORDS PLUS: STEREOCONTROLLED TOTAL SYNTHESIS; ENANTIOSELECTIVE TOTAL SYNTHESIS; OLEFIN METATHESIS APPROACH; CROSS-COUPLING REACTIONS; DRUG DISCOVERY PROCESS; SIDE- CHAIN ANALOGS; EPOTHILONE-B; ASYMMETRIC DIHYDROXYLATION; BIOLOGICAL EVALUATION; 12,13-CYCLOBUTYL EPOTHILONES 13651/ERMOLENKO MS/2002/TETRAHEDRON LETTERS/SYNTHESIS OF EPOTHILONE-B AND EPOTHILONE-D FROM D-GLUCOSE/NO FIELD/KEYWORDS PLUS: ENANTIOSELECTIVE TOTAL SYNTHESIS: OLEFIN METATHESIS APPROACH; STEREOSELECTIVE SYNTHESIS; SORANGIUM- CELLULOSUM; ALDOL CONDENSATION; ANALOGS: (-)-EPOTHILONE-A; 12,13- DESOXYEPOTHILONE-B; DERIVATIVES; DISCOVERY CLUSTER 10 7018/LIST B/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/THE PROLINE-CATALYZED DIRECT ASYMMETRIC 3-COMPONENT MANN1CH REACTION - SCOPE, OPTIMIZATION. AND APPLICATION TO THE HIGHLY ENANTIOSELECTIVE SYNTHESIS OF 1,2-AMINO ALCOHOLS/NO FIELD/KEYWORDS PLUS: LITHIUM ESTER ENOLATE; GLYCOL ALDEHYDE HYDRAZONES; CHIRAL ZIRCONIUM CATALYST; VICINAL AMINO- ALCOHOLS; ALPHA-IMINO ESTERS; BETA-AMINO; ALLIBIS(BINAPHTHOXIDE) COMPLEX; DIASTEREOSELECTIVE SYNTHESIS; UNMODIFIED KETONES; TERNARY COMPLEX 11709/CORDOVA A/2003/TETRAFIEDRON LETTERS/DIRECT ORGANOCATALYTIC ASYMMETRIC MANNICH-TYPE REACTIONS IN AQUEOUS-MEDIA - ONE-POT MANNICH-ALLYLATION REACTIONS/NO FIELD/KEYWORDS PLUS: ALPHA-IMINO ESTERS; ALDOL REACTIONS; DIELS- ALDER; CATALYST; KETONES; COMPLEX 12554/CORDOVA A/2002/TETRAHEDRON LETTERS/ANTI-SELECTIVE SMP-CATALYZED DIRECT ASYMMETRIC MANNICH-TYPE REACTIONS - SYNTHESIS OF FUNCTIONALIZED AMINO-ACID DERIVATIVES/NO FIELD/KEYWORDS PLUS: ALPHA-IMINO ESTERS; ALDOL REACTIONS; KETONES; COMPLEX CLUSTER 11 5691/WU XY/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/HIGHLY ENANTIOSELECTIVE EPOXIDATION OF ALPHA,BETA- UNSATURATED ESTERS BY CHIRAL DIOXIRANE/NO FIELD/KEYWORDS PLUS: CATALYTIC ASYMMETRIC EPOXIDATION; GENERATED IN-SITU; ALPHA,BETA-EPOXY ESTERS; UNFUNCTIONALIZED ALKENES; OPTICAL RESOLUTION; HYDROXY ESTERS; TRANS-OLEFINS; BETA- HYDROXY; KETONES; ACID 7916/SHU LH/2003/JOURNAL OF ORGANIC CHEMISTRY/AN IMPROVED SYNTHESIS OF A KETONE CATALYST FOR ASYMMETRIC EPOXIDATION OF OLEFINS/NO FIELD/KEYWORDS PLUS: GENERATED IN-SITU; HIGHLY ENANTIOSELECTIVE EPOXIDATION; CHIRAL KETONE; UNFUNCTIONALIZED ALKENES: TERMINAL OLEFINS; FLUORO KETONES; CIS-OLEFINS; DIOXIRANES; OXONE; EFFICIENCY 8761/ARMSTRONG A/2002/JOURNAL OF ORGANIC CHEMISTRY/ENANTIOSELECTIVE EPOXIDATION OF ALKENES CATALYZED BY 2- FLUORO-N-CARBETHOXYTROPINONE AND RELATED TROPINONE DERIVATIVES/NO FIELD/KEYWORDS PLUS: GENERATED IN-SITU; MEDIATED ASYMMETRIC EPOXIDATION; CHIRAL KETONE; UNFUNCTIONALIZED OLEFINS; ALPHA- FLUOROCYCLOHEXANONES; DIOXIRANES; ACID; OXONE(R); DIMETHYLDIOXIRANE; CYCLOHEXANONES 181 9194/BORTOLINI O/2002/JOURNAL OF ORGANIC CHEMISTRY/1MPROVED ENANTIOSELECTIVITY IN THE EPOXIDATION OF CINNAMIC ACID-DERIVATIVES W1 TH DIOXIRANES FROM KETO BILE- AC1DS/NO FIELD/KEYWORDS PLUS: CATALYTIC ASYMMETRIC EPOXIDATION; GENERATED IN­ SITU; CHIRAL KETONES; UNFUNCTIONALIZED OLEFINS; DEHYDROCHOLIC ACID; ALKENES; OXIDATIONS; REACTIVITY; GEOMETRY; C-2 9617/TIAN HQ/2002/JOURNAL OF ORGANIC CHEMISTRY/DESIGNING NEW CHIRAL KETONE CATALYSTS - ASYMMETRIC EPOXIDATION OF CIS-OLEFINS AND TERMINAL OLEFINS/NO FIELD/KEYWORDS PLUS: HIGHLY ENANTIOSELECT1VE EPOXIDATION; GENERATED IN-SITU; HYDROGEN-PEROX1DE H2O2; UNFUNCTIONALIZED ALKENES; ABSOLUTE-CONFIGURATION; EPHEDRINE DERIVATIVES; KINETIC RESOLUTION; CONJUGATED DIENES; TRANSITION-STATE; PRIMARY OXIDANT 14233/MATSUMOTO K/2002/TETRAHEDRON LETTERS/CHIRAL KETONE-CATALYZED ASYMMETRIC EPOXIDATION OF OLEFINS WITH OXONE(R)/NO FIELD/KEYWORDS PLUS: GENERATED IN-SITU; UNFUNCTIONALIZED OLEFINS; DIOXIRANES; EFFICIENCY; ALKENES: IMINES CLUSTER 12. 480/HAUSTEDT LO/2003/ANGEWANDTE CHEMIE-INTERNATIONAL ED1TION/THE TOTAL SYNTHESES OF PHORBOXAZOLES - NEW CLASSICS IN NATURAL-PRODUCT SYNTHESIS/AUTHOR KEYWORDS: ANTITUMOR AGENTS; MACROLIDES; PHORBOXAZOLES; SYNTHESIS DESIGN; TOTAL SYNTHESIS/KEYWORDS PLUS: PETASIS-FERRIER REARRANGEMENT: CHIRAL LEWIS­ ACIDS; SPONGE PHORBAS SP; STEREOSELECTIVE SYNTHESIS; ABSOLUTE-CONFIGURATION; MARINE SPONGE; CONVERGENT SYNTHESIS; ASYMMETRIC-SYNTHESIS; ALDOL ADDITIONS; SIDE-CHAIN 742/GONZALEZ MA/2003/ANG EWAN DTE CHEMIE-INTERNATIONAL EDITION/A CONVERGENT TOTAL-SYNTHESIS OF PHORBOXAZOLE-A/AUTHOR KEYWORDS: ANTIFUNGAL AGENTS; ANTITUMOR AGENTS; NATURAL PRODUCTS; OLEFINATION; TOTAL SYNTHESIS/KEYWORDS PLUS; SPONGE PHORBAS SP; ABSOLUTE-CONFIGURATION; MARINE SPONGE; OLEFIN FORMATION; STEREOCHEMISTRY; MACROLIDE; SULFONES; ANALOGS; ESTERS 743/WILLIAMS DR/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/TOTAL-SYNTHESIS OF PHORBOXAZOLE-A/AUTHOR KEYWORDS: ANTITUMOR AGENTS; ASYMMETRIC ALLYLATION; MACROLIDES; NATURAL PRODUCT'S; TOTAL SYNTHESIS/KEYWORDS PLUS: PETASIS-FERRIER REARRANGEMENT: SPONGE PHORBAS SP; STEREOSELECTIVE SYNTHESIS; CONVERGENT SYNTHESIS; MARINE SPONGE; ABSOLUTE-CONFIGURATION: ASYMMETRIC-SYNTHESIS; 1.3- DIOL ACETONIDES; NATURAL-PRODUCTS; SIDE-CHAIN 10188/LI DR/2003/TETRAHEDRON LETTERS/STUDIES ON THE SYNTHESIS OF PHORBOXAZOLE-B - STEREOSELECTIVE- SYNTHESIS OF THE C28-C46 SEGMENT/AUTHOR KEYWORDS: PHORBOXAZOLE B; SYNTHESIS: MUKAIYAMA ALDOL REACTION; OXAZOLE/KEYWORDS PLUS: PETASIS-FERRIER REARRANGEMENT; SPONGE PHORBAS SP; ASYMMETRIC-SYNTHESIS: CONVERGENT SYNTHESIS; NATURAL- PRODUCTS; SIDE-CHAIN; CONSTRUCT ION; OXIDATION; FRAGMENT: SUBUNIT 11032/LIU B/2003/TETRAHEDRON LETTERS/STEREOSELECTIVE-SYNTHESIS OF THE C21-C27 FRAGMENT OF THE PHORBOXAZOLES/NO FIELD/KEYWORDS PLUS; PETASIS-FERRIER REARRANGEMENT; SPONGE PHORBAS SP; ASYMMETRIC-SYNTHESIS; SUBSTITUTED AZEPINONES; CONVERGENT SYNTHESIS; EFFICIENT SYNTHESIS; NATURAL-PRODUCTS; SIDE­ CHAIN; SEGMENT; CONS I'RUCTION 11244/PATERSON I/2003/TE FRAHEDRON LETTERS/TOWARD THE TOTAL-SYNTHESIS OF PHORBOXAZOLE-A - SYNTHESIS OF AN ADVANCED C4-C32 SUBUNIT USING THE JACOBSEN HETERO-DIELS- ALDER REACTION/NO FIELD/KEYWORDS PLUS: PETASIS-FERRIER REARRANGEMENT; SPONGE PHORBAS SP: STEREOSELECTIVE SYNTHESIS; ALDOL REACT IONS; MARINE SPONGE; STEREOCONTROLLED SYNTHESIS; ABSOLUTE-CONFIGURATION; CONVERGENT SYNTHESIS; NATURAL-PRODUCT'S; SIDE-CHAIN 182 CLUSTER 13 1397/TILLACK A/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/ANTI-MARKOVNIKOV HYDROAMINATION OF TERMINAL ALKYNES/AUTHOR KEYWORDS: ALKYNES; HOMOGENEOUS CATALYSIS; HYDROAMINATION; METALLOCENES; TITANIUM/KEYWORDS PLUS: CATALYZED INTERMOLECULAR HYDROAMINATION; OXIDATIVE AMINATION; AROMATIC OLEFINS; UNSATURATED-COMPOUNDS; H ACTIVATION; COMPLEXES; TITANOCENE; ALKENES; FUNCTIONALIZATIONS; DIMETHYLTITANOCENE 2480/ACKERMANN L/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/USE OF GROUP-4 BIS(SULFONAMIDO) COMPLEXES IN THE INTRAMOLECULAR HYDROAMINATION OF ALKYNES AND ALLENES/NO FIELD/KEYWORDS PLUS: CATALYZED INTERMOLECULAR HYDROAMINATION; N-H ACTIVATION; TERMINAL ALKYNES; ETA(1)-PYRROLYL COMPLEXES; SOLVENT PURIFICATION; IMIDO COMPLEXES; AMINES; CYCLIZATION; MECHANISM; SYSTEM 4996/SH1MADA T/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/PALLADIUM- CATALYZED INTERMOLECULAR HYDROAMINATION OF ALKYNES - A DRAMATIC RATE­ ENHANCEMENT EFFECT OF O-AMINOPHENOL/NO FIELD/KEYWORDS PLUS: TERMINAL ALKYNES; COMPLEXES; DIMETHYLTITANOCENE; AMINATION; ALKENES: AMINES 9759/HEUTLING A/2002/JOURNAL OF ORGANIC CHEMISTRY/CP-ASTERISK-2TIME2 - AN IMPROVED CATALYST FOR THE INTERMOLECULAR ADDITION OF N-ALKYL-AMINE AND BENZYLAMINE TO ALKYNES/NO FIELD/KEYWORDS PLUS: ANTI-MARKOVNIKOV-FUNCTIONALIZA FIONS; UNPROTECTED AMINO OLEFINS; 2+2 CYCLOADDITIONS; INTRAMOLECULAR HYDROAMINATION; REGIOSPECIFIC CYCLIZATION; SYNTHETIC APPLICATIONS; UNSATURATED-COMPOUNDS; OXIDATIVE AMINATION; TRANSITION-METALS; AROMATIC OLEFINS 13502/BYTSCHKOV T/2002/TETRAHEDRON LETTERS/THE CP(2)TIME2-CATALYZED INTRAMOLECULAR HYDROAMINATION/ CYCLIZATION OF AMINOALKYNES/AUTHOR KEYWORDS: ALKYNES; AMINATION; AMINES; CATALYSIS; TITANIUM/KEYWORDS PLUS: CATALYZED INTERMOLECULAR HYDROAMINATION; ALKYNE 2+2 CYCLOADDITIONS; SYNTHE11C APPLICATIONS; COMPLEXES; DIMETHYLTITANOCENE; CONCISE; AGENT CLUSTER 14 1417/MOORE DR/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/ELECTRONIC AND STERIC EFFECTS ON CATALYSTS FOR CO2/EPOXIDE POLYMERIZATION - SUBTLE MODIFICATIONS RESULTING IN SUPERIOR ACTIVITIES/AUTHOR KEYWORDS: CARBON DIOXIDE FIXATION; GREEN CHEMISTRY; HOMOGENEOUS CATALYSIS; LIGAND EFFECTS; RING-OPENING POLYMERIZATION/KEYWORDS PLUS: CARBON-DIOXIDE; ALTERNATING COPOLYMERIZATION; CHROMIUM PORPHYRIN; CO2; EPOXIDES; PHENOXIDES; OXIDE: ZINC; DERIVATIVES; RELEVANCE 3252/DARENSBOURG DJ/2003/JOURNAL OF THE AMERICAN CHEMICAL SOC1ETY/COMPARATIVE KINETIC-STUDIES OF THE COPOLYMERIZATION OF CYCLOHEXENE OXIDE AND PROPYLENE­ OXIDE WITH CARBON-DIOXIDE IN THE PRESENCE OF CHROMIUM SALEN DERIVATIVES - IN-SITU FTIR MEASUREMENTS OF COPOLYMER VS CYCLIC CARBONATE PRODUCTION/NO FIELD/KEYWORDS PLUS: ALTERNATING COPOLYMERIZATION; CATALYTIC ACTIVITY: POLYMER SYNTHESIS; EPOXIDES; CO2; COMPLEXES; PHENOXIDES; ZINC; INITIATION; REAGENTS 3645/NAKANO K/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/ASYMMETRIC ALTERNATING COPOLYMERIZATION OF CYCLOHEXENE OXIDE AND CO2 WITH DIMERIC ZINC- COMPLEXES/NO FIELD/KEYWORDS PLUS: CARBON-DIOXIDE; ENANTIOSELECTIVE ADDITION; PRECURSOR CATALYSTS; MECHANISTIC ASPECTS; CHROMIUM PORPHYRIN; MASS- SPECTROMETRY; POLYHYDRIC PHENOL; MAIN-CHAIN; POLYMERIZATION; EPOXIDES 4707/ALLEN SD/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/HIGH-ACTIVITY, SINGLE­ SITE CATALYSTS FOR THE ALTERNATING COPOLYMERIZATION OF CO2 AND PROPYLENE- OXIDE/NO FIELD/KEYWORDS PLUS: CARBON-DIOXIDE; EPOXIDES; POLYMERIZATION; RELEVANCE; COMPLEXES; SYSTEM; ZINC 6107/DARENSBOURG DJ/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/MECHANISTIC ASPECTS OF THE COPOLYMERIZATION REACTION OF CARBON-DIOXIDE AND EPOXIDES, USING A CHIRAL SALEN CHROMIUM CHLORIDE CATALYST/NO FIELD/KEYWORDS PLUS: ALTERNATING COPOLYMERIZATION; CYCLOHEXENE OXIDE; POLYMER SYNTHESIS; COMPLEXES; CO2; PHENOXIDES; ZINC; FIXATION; SITES 183 CLUSTER 15 347/1 lAYASHI Y/2003/ANG EWAN DTE CHEMIE-INTERNATIONAL EDITION/THE DIRECT AND ENANTIOSELECTIVE, ONE-POT, 3-COMPONENT, CROSS- MANNICH REACTION OF ALDEHYDES/AUTHOR KEYWORDS: ALDEHYDES; ASYMMETRIC SYNTHESIS; ENANTIOSELECTIVITY: ORGANOCATALYSTS; 3-COMPONENT REACTION/KEYWORDS PLUS: AMINO-AC1D-DERIVA FIVES; ASYMMETRIC ALDOL REACTIONS; CHIRAL ZIRCONIUM CATALYST; DIELS-ALDER REACTION; ALPHA-IMINO ESTERS; BETA-AMINO; MICHAEL ADDITIONS: UNMODIFIED KETONES; ORGANIC CATALYSIS; ALLIBIS(BINAPHTHOXIDE) COMPLEX 704/.IUHL K/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/THE 1ST ORGANOCATALYTIC ENANTIOSELECTIVE INVERSE-ELECTRON- DEMAND HETERO-DIELS-ALDER REACTION/NO FIELD/KEYWORDS PLUS: ASYMMETRIC ALPHA-AMINATION: 3-COMPONENT MANNICH REACTION: ORGANIC CATALYSIS; MICHAEL ADDITIONS; ALDOL REACTIONS; AMINO-AC1DS; 1.3- D1POLAR CYCLOADDITION; CONJUGATE ADDITION; ALDEHYDES; KETONES 851/HALLAND N/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/HIGHLY ENANTIOSELECTIVE ORGANOCATALYTIC CONJUGATE ADDITION OF MALONATES TO ACYCLIC ALPHA.BETA-UNSATURATED ENONES/AUTHOR KEYWORDS: ASYMMETRIC CATALYSIS; ENONES: KETOESTERS; MALONATES; TETRAHYDROQUINOLINES/KEYWORDS PLUS: ASYMMETRIC MICHAEL ADDITION; BIS(OXAZOLINE) COPPER(II) COMPLEXES; ORGANIC CATALYSIS; ALDOL REACTIONS; L- PROLINE; STRATEGIES; KETONES: ACID; NITROALKANES 7933/MELCHIORRE P/2003/JOURNAL OF ORGANIC CHEMISTRY/DIRECT ENANTIOSELECTIVE MICHAEL ADDI TION OF ALDEHYDES TO VINYL KETONES CATALYZED BY CHIRAL AMINES/NO FIELD/KEYWORDS PLUS: ASYMMETRIC CONJUGATE ADDITION; DIELS-ALDER REACTION; ORGANIC CATALYSIS; ALPHA-AMINATION; ALDOL REACTIONS; L-PROLINE; 1.3-DIPOLAR CYCLOADDITION; 1.4- CONJUGATE ADDITION; ACID; NITROALKANES 8726/HALLAND N/2002/JOURNAL OF ORGANIC CHEMISTRY/ORGANOCATALYTIC ASYMMETRIC CONJUGATE ADDITION OF NITROALKANES TO ALPHA,BETA-UNSATURATED ENONES USING NOVEL IMIDAZOLINE CATALYSTS/NO FIELD/KEYWORDS PLUS: PHASE-TRANSFER CATALYSTS; BOND-FORMING REACTIONS; MICHAEL ADDITIONS; ORGANIC CATALYSIS; CARBONYL­ COMPOUNDS; ALDOL REACTIONS; KETONES; STRATEGIES; ALPHA CLUSTER 16 2509/KIM MJ/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/(S)-SELECTIVE DYNAMIC KINETIC RESOLUTION OF SECONDARY ALCOHOLS BY THE COMBINATION OF SUB1IL1SIN AND AN AMINOCYCLOPENTADIENYLRUTHENIUM COMPLEX AS THE CATALYSTS/NO FIELD/KEYWORDS PLUS: ASYMMETRIC TRANSFORMATIONS: METAL CATALYSIS; CHIRAL ACETATES; EFFICIENT ROUTE; ENOL ACETATES; RACEMIZATION; AMINES; KETONES; ESTERS; ENZYME 8657/KIM MJ/2002/JOURNAL OF ORGANIC CHEMISTRY/ASYMMETRIC TRANSFORMATIONS OF ACYLOXYPHENYL KETONES BY ENZYME-METAL MULTICATALYSIS/NO FIELD/KEYWORDS PLUS: DYNAMIC KINETIC RESOLUTION; SECONDARY ALCOHOLS: HYDROGEN-TRANSFER; CHIRAL ACETALES; ENOL ACETATES; LIPASE; PALLADIUM; RACEMIZATION; ALDEHYDES; AMINES 10449/110 M/2003/TETRAHEDRON LETTERS/RAPID RACEMIZATION OF CHIRAL NON-RACEMIC SEC-ALCOHOLS CATALYZED BY (ETA(5)-C-5(CH3)(5))RU COMPLEXES BEARING TERTIARY PHOSPHINE-PRIMARY AMINE CHELATE LIGANDS/NO FIELD/KEYWORDS PLUS: DYNAMIC KINETIC RESOLUTION; ASYMMETRIC HYDROGEN- TRANSFER; SECONDARY ALCOHOLS; ENZYMATIC RESOLUTION; ENOL ACETATES; RUTHENIUM; KETONES; MECHANISM; DIOLS; ROUTE 13672/RUNMO ABL/2002/TETRAHEDRON LETTERS/DYNAMIC KINETIC RESOLU! ION OF GAMMA­ HYDROXY ACID-DERIVATIVES/NO FIELD/KEYWORDS PLUS: RUTHEN1UM-CATALYZED RACEMIZATION: CHIRAL BUILDING-BLOCKS; ENZYMATIC RESOLUTION; ENANTIOSELECTIVE SYNTHESIS: SECONDARY ALCOHOLS; ORGANIC-SOLVENTS; AMINO ALCOHOLS; LIPASE; LACTONES; COMPLEXES 184 CLUSTER 17 1763/HULTZSCH KC/2002/ANGEWANDTE CHEMIE-INl'ERNATIONAL EDITION/THE 1ST POLYMER- SUPPORTED AND RECYCLABLE CHIRAL CATALYST FOR ENANTIOSELECTIVE OLEFIN METATHESIS/AUTHOR KEYWORDS: ASYMMETRIC CATALYSIS; IMMOBILIZATION; METATHESIS; MOLYBDENUM; SOLID-PHASE SYNTHESIS/KEYWORDS PLUS: RING-CLOSING METATHESIS; KINETIC RESOLUTION; COMPLEXES; KETONES; DERIVATIVES; ALKENES 2373/VANVELDHU1ZEN JJ/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/CH1RAL RU- BASED COMPLEXES FOR ASYMMETRIC OLEFIN METATHESIS - ENHANCEMENT OF CATALYST ACTIVITY THROUGH STERIC AND ELECTRONIC MODIFICATIONS/NO FIELD/KEYWORDS PLUS: RING-CLOSING METATHESIS; OPENING-CROSS METATHESIS; ENANTIOSELECTIVE SYNTHESIS; ORGANIC-SYNTHESIS; UNSATURATED ALCOHOLS; RUTHENIUM CARBENE; TERTIARY ETHERS; STYRENYL ETHERS; EFFICIENT; ACRYLONITRILE 4114/TSANG WCP/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/AN ENANTIOMERICALLY PURE ADAMANTYLIMIDO MOLYBDENUM ALKYLIDENE COMPLEX - AN EFFECTIVE NEW CATALYST FOR ENANTIOSELECTIVE OLEFIN METATHESIS/NO FIELD/KEYWORDS PLUS: RING-CLOSING METATHESIS; CYCLIC TERTIARY ETHERS; LIGANDS; POLYMERIZATION 5370/TENG X/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/ENHANCEMENT OF ENANTIOSELECTIV1TY BY THF IN ASYMMETRIC MO- CATALYZED OLEFIN METATHESIS. CATALYTIC ENANTIOSELECTIVE SYNTHESIS OF CYCLIC TERTIARY ETHERS AND SPIROCYCLES/NO FIELD/KEYWORDS PLUS: RING-CLOSING METATHESIS; 1MID0 ALKYLIDENE COMPLEXES; CHIRAL ZIRCONIUM CATALYST; MANNICH-TYPE REACTIONS; KINETIC RESOLUTION; KETONES; IMINES; ACID 6005/DOLMAN SJ/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/EFFIC1ENT CATALYTIC ENANTIOSELECTIVE SYNTHESIS OF UNSATURATED AMINES - PREPARATION OF’ SMALL-RING AND MEDIUM- RING CYCLIC AMINES THROUGH MO-CATALYZED ASYMMETRIC RING- CLOSING METATHESIS IN THE ABSENCE OF SOLVENT/NO FIELD/KEYWORDS PLUS: OLEFIN METATHESIS; KINETIC RESOLUTION; STYRENYL ETHERS; HETEROCYCLES; CHROMENES; MECHANISM; ALKALOIDS; COMPLEXES 6288/VANVELDHUIZEN JJ/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/A RECYCLABLE CHIRAL RU CATALYST FOR ENANTIOSELECTIVE OLEFIN METATHESIS - EFFICIEN T CATALYTIC ASYMMETRIC RING-OPENING/ CROSS METATHESIS IN AIR/NO FIELD/KEYWORDS PLUS: CROSS­ METATHESIS; CLOSING METATHESIS; MECHANISM 6608/KIELY AF/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/ENANTIOSELECTIVE SYNTHESIS OF MEDIUM-RING HETEROCYCLES. TERTIARY ETHERS. AND TERTIARY ALCOHOLS BY MO-CATALYZED RING- CLOSING METATHESIS/NO FIELD/KEYWORDS PLUS: KINETIC RESOLUTION: KETONES CLUSTER 18 1508/TAN DS/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/SYNTHES1S OF THE FUNCTIONALIZED TRICYCLIC SKELETON OF GUANACASTEPENE-A - A TANDEM EPOXIDE- OPENING BETA-ELIMINATION/ KNOEVENAGEL CYCLIZATION/NO FIELD/KEYWORDS PLUS: RING­ SYSTEM; 3-OXO-4-PENTENOATE; CONVERSION; ALDEHYDES; FUNGUS 1509/LIN SN/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDIT1ON/A STEREOSELECTIVE ROUTE TO GUANACASTEPENE-A THROUGH A SURPRISING EPOXIDATION/NO FIELD/KEYWORDS PLUS: BIOLOGICAL-ACTIVITY; ALCOHOLS; CONVERSION; ALDEHYDES; REDUCTION; KETONES; FUNGUS; ALPHA 11164/MEHTA G/2003/TETRAHEDRON LETTERS/GUANACASTEPENE-A TOTAL-SYNTHESIS - CONSTRUCTION OF THE TRICYCLIC ISO-GUANACASTEPANE. EPI-GUANACASTEPANE AND GUANACASTEPANE FRAMEWORKS/NO FIELD/KEYWORDS PLUS: CARBON SKELETON; RING- SYSTEM; CORE; CYCLIZATION; PROGRESS; PORTION; FUNGUS 185 12141/MAGNUS P/2002/TETRAHEDRON LETTERS/SYNTHESIS OF THE HYDROAZULENE PORTION OF GUANACASTEPENE-A USING A (2.3)SIGMATROP1C SULFOXIDE REARRANGEMENT - OBSERVATIONS ON SILYL ENOL ETHER ELECTROPHILIC CHEMISTRY FOR THE INTRODUCTION OF THE C-13 HYDROXYL GROUP/AUTHOR KEYWORDS: GUANACASTEPENE; (2.3)SIGMATROPIC REARRANGEMENT; SILYL ENOL ETHERS/KEYWORDS PLUS: M-CHLOROPERBENZOIC ACID: STEREOSELECTIVE SYNTHESIS: ALPHA-HYDROXY; OXIDATION; FUNGUS 12579/BOYER FD/2002/TETRAHEDRON LETTERS/SYNTHESIS OF A HIGHLY FUNCTIONALIZED TRICYCLIC RING-SYSTEM RELATED TO GUANACASTEPENE VIA A TANDEM RING-CLOSING METATHESIS REACTION/NO FIELD/KEYWORDS PLUS: OLEFIN METATHESIS; D1ENYNES; CONSTRUCTION; CATALYSTS: ALCOHOLS; EPOXIDES; FUNGUS 12714/MEHTA G/2002/TETRAHEDRON LETTERS/TOWARDS A TOTAL SYNTHESIS OF GUANACASTEPENE-A - CONSTRUCTION OF FULLY FUNCTIONALIZED AB AND BC RING SEGMENTS/NO FIELD/KEYWORDS PLUS: DOLASTANE DITERPENES; FUNGUS 13041/DUDLEY GB/2002/TETRAHEDRON LETTERS/ON THE USE OF DEUTERIUM-ISOTOPE EFFECTS IN CHEMICAL SYNTHESIS/NO FIELD/KEYWORDS PLUS: GUANACASTEPENE; FUNGUS 13407/NGUYEN TM/2002/TETRAHEDRON LETTERS/PROGRESS TOWARDS THE TOTAL SYNTHESIS OF GUANACASTEPENE-A - APPROACHES TO THE CONSTRUCTION OF QUATERNARY CARBONS AND THE 5-7-6-TRICYCLIC CARBON SKELETON/NO FIELD/KEYWORDS PLUS: C-H INSERTION; CYCLIZATION; FUNGUS CLUSTER 19 9986/VOGL EM/2002/JOURNAL OF ORGANIC CHEMISTRY/PALLADIUM-CATALYZED MONOARYLATION OF NITROALKANES/NO FIELD/KEYWORDS PLUS: ALPHA-ARYLATION; KETONES; LIGANDS 110156/LIU P/2003/TETRAHEDRON LETTERS/A HIGHLY-ACTIVE CATALYST SYSTEM FOR THE HETEROARYLATION OF ACETONE/NO FIELD/KEYWORDS PLUS: ALPHA-ARYLATION; ASYMMETRIC ARYLATION; PALLADIUM; KETONES: ESTERS 13761/KASHIN AN/2002/TETRAHEDRON LETTERS/PALLADIUM-CATALYZED ARYLATION OF SULFONYL CH-AC1DS/AUTHOR KEYWORDS; PALLADIUM; SULFONES; CH-ACIDS; CATALYSIS; ARYL HALIDES; ARYLATION; CARBANIONS/KEYWORDS PLUS: ALPHA-ARYLATION: ARYL HALIDES; KETONES; COMPLEX 14370/TERAO Y/2002/TETRAHEDRON LETTERS/PALLADIUM-CATALYZED ALPHA-ARYLATION OF ALDEHYDES WITH ARYL BROMIDES/AUTHOR KEYWORDS: ARYL BALIDES; ARYLATION ALDEHYDES; PALLADIUM AND COMPOUNDS/KEYWORDS PLUS: ALPHA,BETA-UNSATURATED CARBONYL; REGIOSELECTIVE ARYLATION: KE TONES; HALIDES; NAPHTHOLS CLUSTER 20 11271/CHAN DMT/2003/TETRAHEDRON LETTERS/COPPER PROMOTED C-N AND C-0 BOND CROSS­ COUPLING WITH PHENYL AND PYRIDYLBORONATES/NO FIELD/KEYWORDS PLUS: ARYLBORONIC ACIDS; ROOM-TEMPERATURE; CUPRIC ACETATE; BORONIC ACIDS; DIARYL ETHERS; PHENYLBORONIC ACIDS; ARYLATION; ARYL; IMIDAZOLES; PHENOLS 11762/LAM PYS/2003/TETRAHEDRON LETTERS/N-ARYLATION OF ALPHA-AMINOESTERS WITH P- TOLYLBORONIC ACID- PROMOTED BY COPPER(II) ACETATE/NO FIELD/KEYWORDS PLUS: CROSS­ COUPLING REACTIONS; ARYLBORONIC ACIDS; C-N; PHENYLBORONIC ACIDS; CUPRIC ACETATE; ROOM-TEMPERATURE; BORONIC ACIDS; DIARYL ETHERS; O-ARYLA'I ION; AMINO-ACIDS 13699/LAM PYS/2002/TETRAIIEDRON LETTERS/COPPER-PROMOTED C-N BOND CROSS-COUPLING WITH PHENYLSTANNANE/NO FIELD/KEYWORDS PLUS: ARYLBORONIC ACIDS; PHENYLBORONIC ACIDS; ROOM- TEMPERATURE; CUPRIC ACETATE; BORONIC ACIDS; DIARYL ETHERS; O- ARYLATION; PHENOLS; AMINES: IMIDAZOLES 186 CLUSTER 21 1376/KITA T/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/C-2-SYMMETRIC CHIRAL PENTACYCLIC GUANIDINE - A PHASE- TRANSFER CATALYST FOR THE ASYMMETRIC ALKYLATION OF TERT- BUTYL GLYCINATE SCHIFF-BASE/AUTHOR KEYWORDS: ALKYLATION; ASYMMETRIC SYNTHESIS; PHASE- TRANSFER CATALYSIS; SYNTHETIC METHODS/KEYWORDS PLUS: ALPHA-AMINO-ACIDS; QUATERNARY AMMONIUM SALT; SPONGE CRAMBE-CRAMBE; ENANTIOSELECTIVE SYNTHESIS; S TEREOSELECTIVE SYNTHESIS; BICYCLIC GUANIDINE; MICHAEL REACTION; PTILOMYCALIN-A; AMIDINIUM IONS; ALKALOIDS 10175/ALLINGHAM MT/2003/TETRAHEDRON LETTERS/SYNTHESIS AND APPLICATIONS OF C-2- SYMMETRIC GUANIDINE BASES/AUTHOR KEYWORDS: PHASE TRANSFER; CATALYSIS; GUANIDINE/KEYWORDS PLUS: PHASE-TRANSFER CATALYSIS; ALPHA-AMINO-ACIDS; ENANTIOSELECTIVE SYNTHESIS; MICHAEL REACTION: CINCHONA ALKALOIDS; ALKYLATION; ESTERS 12125/ARAI S/2002/TETRAHEDRON LETTERS/PHASE-TRANSFER-CATALYZED ASYMMETRIC MICHAEL REACTION USING NEWLY-PREPARED CHIRAL QUATERNARY AMMONIUM-SALTS DERIVED FROM L-TARTRATE/AUTHOR KEYWORDS: ASYMMETRIC; CATALYSIS; MICHAEL REACTION; PHASE-TRANSFER/KEYWORDS PLUS: ALPHA-AMINO-ACIDS; ENANTIOSELECTIVE SYNTHESIS; LIGANDS CLUSTER 22 779/DUTHALER RO/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDIT1ON/PROL1NE-CATALYZED ASYMMETRIC ALPHA-AMINATION OF ALDEHYDES AND KETONES - AN ASTONISHINGLY SIMPLE ACCESS TO OPTICALLY- ACTIVE ALPHA-HYDRAZINO CARBONYL-COMPOUNDS/AUTHOR KEYWORDS: AMINATION; AZODICARBOXYLATES; CATALYSIS; ENANTIOSELECTIVITY; PROLINE/KEYWORDS PLUS: 3-COMPONENT MANNICH REACTION; ALDOL REACTIONS; AMINO­ ACIDS; MICHAEL ADDITIONS; ENANTIOSELECTIVE AMINATION; ARYLGLYCINES; COMPLEXES; ALCOHOLS; ROUTE 3605/TANG Z/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/NOVEL SMALL ORGANIC- MOLECULES FOR A HIGHLY ENANTIOSELECTIVE DIRECT ALDOL REACTION/NO FIELD/KEYWORDS PLUS: ASYMMETRIC ALPHA-AMINATION: AMINO-AC1DS; ATOM ECONOMY; L- PROLINE; ALDEHYDES; CATALYSTS; KETONES; COMPLEX; ROUTE 4113/HARADA S/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/DIRECT CATALYTIC ASYMMETRIC MICHAEL REACTION OF HYDROXYKETONES - ASYMMETRIC ZN CATALYSIS WITH A ET2ZN/L1NKED- BINOL COMPLEX/NO FIELD/KEYWORDS PLUS: LINKED-BINOL COMPLEX; MANNICH-TYPE REACTIONS; BIS(OXAZOLINE) COPPER(II) COMPLEXES; QUATERNARY AMMONIUM SALT; AMINO ACID-DERIVAFIVES; SILYL ENOL ETHERS; ALDOL REACTION; UNMODIFIED KETONES; ALPHA-AMINO; ALLIBIS(BINAPHTHOXIDE) COMPLEX CLUSTER 23 2207/ZHONG XH/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/ALLOYED ZNXCD1-XS NANOCRYSTALS WITH HIGHLY NARROW LUMINESCENCE SPECTRAL WIDTH/NO FIELD/KEYWORDS PLUS: LIGHT-EMITTING-DIODES; QUANTUM DOTS; SEMICONDUCTOR CLUSTERS; CDSE NANOCRYSTALS; SIZE; ZNS; NANOPARTICLES; CDXZN1-XS; POLYMER; ENERGY 2381/LI J.I/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/LARGE-SCALE SYNTHESIS OF NEARLY MONODISPERSE CDSE/CDS CORE/ SHELL NANOCRYSTALS USING AIR-STABLE REAGENTS VIA SUCCESSIVE ION LAYER ADSORPTION AND REACTION/NO FIELD/KEYWORDS PLUS: LIGHT-EMITTING-DIODES; SHELL QUANTUM DOTS; SEMICONDUCTOR NANOCRYSTALS; ALTERNATIVE ROUTES; EPITAXIAL- GROWTH; CLUSTERS; ZNSE; ELECTROLUMINESCENCE; NUCLEATION; DEPOSITION 3086/ZHONG XH/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/COMPOSITION-TUNABLE ZNXCD1-XSE NANOCRYSTALS WITH HIGH LUMINESCENCE AND STABILITY/NO FIELD/KEYWORDS PLUS: LIGHT-EMITTING-DIODES; QUANTUM DOTS; CORE/ SHELL NANOCRYSTALS; CDSE NANOCRYSTALS; EPITAXIAL-GROWTH; SIZE; CORE; SEMICONDUCTORS; NANOPARTICLES; CDXZN1-XS 187 3910/GUO WH/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/LUMINESCENT CDSE/CDS CORE/SHELL NANOCRYSTALS IN DENDRON BOXES - SUPERIOR CHEMICAL, PHOTOCHEMICAL AND THERMAL- STABILITY/NO FIELD/KEYWORDS PLUS: QUANTUM DOTS; SEMICONDUCTOR CLUSTERS; EPITAXIAL-GROWTH; CORED DENDRIMERS; MONODISPERSE; PHOTOLUMINESCENCE; NANOPARTICLES; COMPOSITES; PRECURSOR; NANORODS 6823/QU LFI/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/CONTROL OF PHOTOLUMINESCENCE PROPERTIES OF CDSE NANOCRYSTALS IN GROWTH/NO FIELD/KEYWORDS PLUS: QUANTUM DOTS; SEMICONDUCTOR CLUSTERS; II-VI; EMISSION CLUSTER 24 3357/STAUFFER SR/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/FLUORESCENCE RESONANCE ENERGY-TRANSFER (FRET) AS A HIGH- THROUGHPUT ASSAY FOR COUPLING REACTIONS - ARYLATION OF AMINES AS A CASE-STUDY/NO FIELD/KEYWORDS PLUS: PALLADIUM-CATALYZED AMINATION; ARYL CHLORIDES; HOMOGENEOUS CATALYSTS; MASS- SPECTROMETRY; ARYLBORONIC ACIDS; BOND FORMATION; C-N; ENANTIOSELECTIVE CATALYSTS; CROSS- COUPLINGS; MILD CONDITIONS 8536/URGAONKAR S/2003/JOURNAL OF ORGANIC CHEMISTRY/P(1-BUNCH2CH2)(3)N - AN EFFECTIVE LIGAND IN THE PALLADIUM- CATALYZED AMINATION OF ARYL BROMIDES AND 1ODIDES/NO FIELD/KEYWORDS PLUS: ROOM-TEMPERATURE AMINATION; PHOSPHORUS­ NITROGEN BOND; CRYSTAL-STRUCTURE; SECONDARY-AMINES; CHLORIDES; HALIDES; SYSTEM; COMPLEXES; TRIS(MORPHOLINO)PHOSPHINE; PROAZAPHOSPHATRANES 8568/MARGOLIS B.I/2003/JOURNAL OF ORGANIC CHEMISTRY/AN EFFICIENT ASSEMBLY OF HETEROBENZAZEPINE RING-SYSTEMS UTILIZING AN INTRAMOLECULAR PALLADIUM- CATALYZED CYCLOAMINATION/NO FIELD/KEYWORDS PLUS: NITROGEN BOND FORMATION; ARYL HALIDES; AMINATION; CHLORIDES; BROMIDES; SCOPE CLUSTER 25 1703/LEWIS FD/2002/ANGE WANDTE CHEMIE-INTERNATIONAL EDITION/DYNAMICS AND ENERGETICS OF HOLE TRAPPING IN DNA BY 7- DEAZAGUANINE/AUTHOR KEYWORDS: DNA CONJUGATES; ELECTRON TRANSFER; PHOTOOXIDATION; PICOSECOND SPECTROSCOPY/KEYWORDS PLUS: PHOTOINDUCED ELECTRON-TRANSFER; DUPLEX DNA; TRANSPORT; OXIDATION; DISTANCE; SEQUENCE; HAIRPINS; 8- OXOGUANINE; DEPENDENCE; STACKING 2025/BELJONNE D/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/PATHWAYS FOR PHOTOINDUCED CHARGE SEPARATION IN DNA HAIRPINS/NO FIELD/KEYWORDS PLUS: ELECTRON-TRANSFER; HOLE TRANSPORT; DISTANCE DEPENDENCE; MIGRATION; MOLECULES; MECHANISM; PROTEINS; DYNAMICS; SEQUENCE; LINKERS 3739/LEW1S FD/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/DYNAMICS AND ENERGETICS OF SINGLE-STEP HOLE TRANSPORT IN DNA HAIRPINS/NO FIELD/KEYWORDS PLUS: DISTANCE CHARGE-TRANSPORT; PHOTOINDUCED ELECTRON-TRANSFER; DOUBLE-HELICAL DNA; B-FORM DNA; LONG- RANGE; TRANSIENT ABSORPTION; HOPPING MECHANISM; RADICAL­ CATION; DEPENDENCE; OXIDATION 4779/LEWIS FD/2002/JOURNAL OF THE AMERICAN CHEMICAL SOC1ETY/FORMATION AND DECAY OF LOCALIZED CONTACT RADICAL-ION PAIRS IN DNA HAIRPINS/NO FIELD/KEYWORDS PLUS: PHOTOINDUCED ELECTRON-TRANSFER; CHARGE- TRANSFER; HOLE-TRANSPORT; DISTANCE DEPENDENCE; DYNAMICS; OLIGONUCLEOTIDES; DERIVATIVES; ENERGETICS; LINKERS; BASES CLUSTER 26 1117/BEDFORD RB/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/SIMPLE MIXED TRICYCLOHEXYLPHOSPHANE-TRIARYLPHOSPHITE COMPLEXES AS EXTREMELY HIGH- ACTIVITY CATALYSTS FOR THE SUZUKI COUPLING OF ARYL CHLORIDES/AUTHOR KEYWORDS: C-C COUPLING; HOMOGENEOUS CATALYSIS; PALLADIUM; SUZUKI REACTION/KEYWORDS PLUS: ARYLBORONIC ACIDS; PHENYLBORONIC ACID; CROSS- COUPLINGS 188 1844/BOTELLA L/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/A CONVENIENT OXIME- CARBAPALLADACYCLE-CATALYZED SUZUKI CROSS- COUPLING OF ARYL CHLORIDES IN WATER/AUTHOR KEYWORDS: BIARYLS; C-C-COUPLING; CROSS-COUPLING; PALLADIUM; PHASE­ TRANSFER CATALYSIS/KEYWORDS PLUS: N-HETEROCYCLIC CARBENES; HECK-TYPE REACTIONS; ARYLBORONIC ACIDS; PHENYLBORONIC ACID; COMPLEXES; HALIDES; PALLADACYCLES; PALLADIUM(II); CHLOROARENES; PRECURSORS 9166/ALONSO DA/2002/JOURNAL OF ORGANIC CHEM1STRY/HIGHLY-ACTIVE OXIME-DERIVED PALLADACYCLE COMPLEXES FOR SUZUKI-MIYAURA AND ULLMANN-TYPE COUPLING REACTIONS/NO FIELD/KEYWORDS PLUS: EFFICIENT CATALYST PRECURSORS; UNACTIVATED METHYL-GROUPS; ARYLBORONIC ACIDS; ARYL CHLORIDES; CROSS- COUPLINGS; C-C; STRUCTURAL CHARACTERIZATION; LIGANDLESS PALLADIUM; SYMMETRICAL BIARYLS: ROOM­ TEMPERATURE 10348/TAO B/2003/TETRAHEDRON LETTERS/TRANS-PD(OAC)2(CY2NH)(2) CATALYZED SUZUKI COUPLING REACTIONS AND ITS TEMPERATURE-DEPENDENT ACTIVITIES TOWARD ARYL BROMIDES/NO FIELD/KEYWORDS PLUS: HIGHLY-ACTIVE CATALYSTS; N-HETEROCYCLIC CARBENES; ARYLBORONIC ACIDS; PALLADIUM CATALYSTS; C-C; CHLORIDES; COMPLEXES; WATER; CONVENIENT; LIGAND 13195/TAO B/2002/TETRAHEDRON LETTERS/PD(OAC)(2)/2-ARYL-2-OXAZOLINES CATALYZED SUZUKI COUPLING REACTIONS OF ARYL BROMIDES AND ARYLBORONIC ACIDS/NO FIELD/KEYWORDS PLUS: N-HETEROCYCLIC CARBENES; HIGHLY-ACTIVE CATALYSTS; HECK TYPE REACTIONS; EFFICIENT CATALYSTS; COMPLEXES; CHLORIDES; PALLADACYCLES; DERIVATIVES; CONVENIENT; VANCOMYCIN CLUSTER 27 384/YAO QW/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/OLEFIN METATHESIS IN THE IONIC LIQUID l-BUTYL-3- METHYLIMIDAZOLIUM HEXAFLUOROPHOSPHATE USING A RECYCLABLE RU CATALYST - REMARKABLE EFFECT OF A DESIGNER IONIC TAG/AUTHOR KEYWORDS: CARBENE COMPLEXES: HOMOGENEOUS CATALYSIS; IONIC LIQUIDS; METATHESIS; RUTHENIUM/KEYWORDS PLUS: ENYNE METATHESIS; EFFICIENT; COMPLEXES; LIGANDS; MECHANISM 1142/CONNON SJ/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/A SELF-GENERATING, HIGHLY-ACTIVE, AND RECYCLABLE OLERIN- METATHESIS CATALYST/AUTHOR KEYWORDS: ALKENES; HOMOGENEOUS CATALYSIS; METATHESIS; POLYMERS; RUTHENIUM/KEYWORDS PLUS: RING-CLOSING METATHESIS; OLEFIN-METATHESIS; IMIDAZOL1N-2-YLIDENE LIGANDS; MINIMAL PURIFICATION; CARBENE COMPLEXES: ORGANIC-SYNTHESIS; RUTHENIUM; POLYSTYRENE; EFFICIEN T; PRODUCT S 12227/GRELA K/2002/TETRAHEDRON LETTERS/A PS-DES IMMOBILIZED RUTHENIUM CARBENE - A ROBUST AND EASILY RECYCLABLE CATALYST FOR OLEFIN METATHESIS/NO FIELD/KEYWORDS PLUS: CONVENIENT METHOD; ORGANIC-SYNTHESIS; CARBON- DIOXIDE; EFFICIENT; SUPPORTS CLUSTER 28 879/CARDENAS DJ/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDIT1ON/ADVANCES IN FUNCTIONAL-GROUP-TOLERANT METAL-CATALYZED ALKYL- ALKYL CROSS-COUPLING REACTIONS/AUTHOR KEYWORDS: ALKANES; BORANES; CROSS-COUPLING; NICKEL: PALLADIUM/KEYWORDS PLUS: STILLE REACTION: ARYL CHLORIDES; EFFICIENT; MECHANISM; BROMIDES; CENTERS 1123/TSUJI T/2002/ANG EWAN DTE CHEMIE-INTERNATIONAL EDITION/COBALT-CATALYZED COUPLING REACTION OF ALKYL-HALIDES WITH ALLYLIC GRIGNARD-REAGENTS/AUTHOR KEYWORDS: ALLYLATION; C-C COUPLING; COBALT; CROSS- COUPLING; RADICAL REACT1ONS/KEYWORDS PLUS: CARBON CENTERS; BETA-HYDROGENS; EFFICIENT; ALKENYLATION 1168/NETHERTON MR/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/SUZUKI CROSS­ COUPLINGS OF ALKYL TOSYLATES THAT POSSESS BETA- HYDROGEN ATOMS - SYNTHETIC AND MECHANISTIC STUDIES/AUTHOR KEYWORDS: ALKYL TOSYLATES; C-C COUPLING: LIGAND EFFECTS; PALLADIUM; SUZUKI REACTION/KEYWORDS PLUS: BONDS 189 1539/KlRCHHOFF JH/2002/ANGE WANDTE CHEMIE-INTERNATIONAL EDITION/A METHOD FOR PALLADIUM-CATALYZED CROSS-COUPLINGS OF SIMPLE ALKYL CHLORIDES - SUZUKI REACTIONS CATALYZED BY (PD- 2(DBA)(3))/PCY3/AUTHOR KEYWORDS: ALKYL CHLORIDES: CROSS-COUPLING; HOMOGENEOUS CATALYSIS; PALLADIUM; P L1GANDS/KEYWORDS PLUS: BETA-HYDROGENS: EFFICIENT; CENTERS 4813/K1RCHHOFF JH/2002/.IOURNAL OF THE AMERICAN CHEMICAL SOCIETY/BORONIC ACIDS - NEW COUPLING PAR TNERS IN ROOM-TEMPERATURE SUZUKI REACTIONS OF ALKYL BROMIDES - CRYSTALLOGRAPHIC CHARACTERIZATION OF AN OXIDATIVE-ADDITION ADDUCT GENERATED UNDER REMARKABLY MILD CONDITIONS/NO FIELD/KEYWORDS PLUS: BETA-HYDROGENS CLUSTER 29 9928/CARR1GAN MD/2002/JOURNAL OF ORGANIC CHEMISTRY/A SIMPLE AND EFFICIENT CHEMOSELECTIVE METHOD FOR THE CATALYTIC DEPROTECTION OF ACETALS AND KETALS USING BISMUTH TRIFLATE/NO FIELD/KEYWORDS PLUS: SELECTIVE CLEAVAGE; HIGHLY EFFICIENT; TRIFLUOROMETHANESULFONATE; CHLORIDE; ACYLATION 10364/PETERSON KE/2003/TETRAHEDRON LETTERS/BISMUTH COMPOUNDS IN ORGANIC- SYNTHESIS - SYNTHESIS OF RESORCINARENES USING BISMUTH TRIFLATE/AUTHOR KEYWORDS: BISMUTH AND COMPOUNDS; RESORCINARENES; LEWIS ACIDS; ENVIRONMENT-FRIENDLY CATALYSTS/KEYWORDS PLUS: COLUMNAR LIQUID-CRYSTALS; FRIEDEL-CRAFTS ACYLATION; HOST GUEST COMPLEXATION; DIELS-ALDER REACTION; EFFICIENT METHOD; ALDEHYDES; TRIFLUOROMETHANESULFONATE; CHLORIDE; KETONES; CALIX(4)RESORCINARENES 10703/REDDY AV/2003/TETRAHEDRON LETTERS/BISMUTH TRIFLATE CATALYZED CONJUGATE ADDITION OF INDOLES TO ALPHA.BETA-ENONES/AUTHOR KEYWORDS: BISMUTH TRIFLATE; INDOLE; ALPHA.BETA-ENONES; ADDITION REACT1ONS/KEYWORDS PLUS: ELECTRON-DEFICIENT OLEFINS; DIELS-ALDER REACTION; EFFICIENT METHOD; HAPALOSIPHON-FONTINALIS; TRIFLUOROMETHANESULFONATE: HAPALINDOLES: ALDEHYDES; ALKALOIDS; CHLORIDE; KETONES CLUSTER 30 672/CHOI TL/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/CONTROLLED LIVING RING­ OPENING-METATHESIS POLYMERIZATION BY A FAST-INITIATING RUTHENIUM CATALYST/AUTHOR KEYWORDS: COPOLYMERIZATION; METATHESIS; N LIGANDS; RING­ OPENING POLYMERIZATION; RUTHENIUM/KEYWORDS PLUS: OLEFIN CROSS-METATHESIS; N- HETEROCYCLIC CARBENES; ALKYLIDENE COMPLEXES; COPOLYMERS; LIGANDS; ROMP; NORBORNENES; MECHANISM 1089/LOVE JA/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/A PRACTICAL AND HIGHLY-ACTIVE RUTHENIUM-BASED CATALYST THAT EFFECTS THE CROSS METATHESIS OF ACRYLONITRILE/NO FIELD/KEYWORDS PLUS: OLEFIN METATHESIS; LIGANDS; EFFICIENT; VINYLPHOSPHONATE; MECHANISM; ALKENES; COMPLEX 1143/CHOI TL/2002/ANG EWAN DTE CHEMIE-INTERNATIONAL EDITION/SYNTHESIS OF A,B- ALTERNATING COPOLYMERS BY RING-OPENING- INSERTION-METATHESIS POLYMERIZATION/AUTHOR KEYWORDS: COPOLYMERIZATION; CROSS-COUPLING; METATHESIS; RING-OPENING POLYMERIZATION; RUTHENIUM/KEYWORDS PLUS: OLEFIN CROSS-METATHESIS; CATALYSTS; COMPLEXES; LIGANDS; ALKENES 1260/CHATTERJEE AK/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/FORMAL VINYL C- H ACTIVATION AND ALLYLIC OXIDATION BY OLEFIN METATHESIS/NO FIELD/KEYWORDS PLUS: CROSS-METATHESIS; CATALYSTS; ALKYNES CLUSTER 31 1090/GRELA K/2002/ANG EWAN DTE CHEMIE-INTERNATIONAL EDITION/A HIGHLY EFFICIENT RUTHENIUM CATALYST FOR METATHESIS REACTION/NO FIELD/KEYWORDS PLUS: SELECTIVE CROSS-METATHESIS; OLEFIN METATHESIS; STABLE CARBENES; MECHANISM: BEARING; LIGANDS 1463/WAKAMATSU H/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/A NEW HIGHLY EFFICIENT RUTHENIUM METATHESIS CATALYST/NO FIELD/KEYWORDS PLUS: OLEFIN METATHESIS; CROSS METATHESIS; LIGANDS; COMPLEX 190 1734/WAKAMATSU H/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/A HIGHLY-ACTIVE AND AIR-STABLE RUTHENIUM COMPLEX FOR OLEFIN METATHESIS/AUTHOR KEYWORDS: BIARYLS; CARBENE LIGANDS; METATHESIS; O LIGANDS; RUTHENIUM/KEYWORDS PLUS: CATALYSTS 11514/DUNNE AM/2003/TETRAHEDRON LETTERS/A HIGHLY EFFICIENT OLEFIN METATHESIS INITIATOR - IMPROVED SYNTHESIS AND REACTIVITY STUDIES/NO FIELD/KEYWORDS PLUS: OPENING-CROSS-METATHESIS; IMIDAZOLIN-2- YLIDENE LIGANDS; RUTHENIUM COMPLEX: ORGANIC-SYNTHESIS; CATALYSTS; CYCLOALKENES; DERIVATIVES; GENERATION; CHEMISTRY CLUSTER 32 5089/HAYASHI T/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/RHODIUM-CATALYZED ASYMMETRIC 1,4-ADDITION OF ARYLTITANIUM REAGENTS GENERATING CHIRAL TITANIUM ENOLATES - ISOLATION AS SILYL ENOL ETHERS/NO FIELD/KEYWORDS PLUS: CONJUGATE ADDITION; ARYLBORONIC ACIDS; ALPHA.BETA-UNSATURATED ESTERS; ORGANOBORONIC ACIDS; ALDOL REACTION; OLEFINS; KETONES; ENONES 5305/YOSHIDA K/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/A NEW-TYPE OF CATALYTIC TANDEM 1,4-ADDIl ION-ALDOL REACTION WHICH PROCEEDS THROUGH AN (OXA-PI- ALLYL)RHODIUM 1NTERMEDIATE/NO FIELD/KEYWORDS PLUS: REDUCTIVE ALDOL REACTION: ASYMMETRIC 1.4- ADDITION; CONJUGATE ADDITION; ARYLBORONIC ACIDS; ALPHA.BETA- UNSATURATED ESTERS: ORGANOBORONIC ACIDS; RHODIUM; REAGENTS; KETONES; ENONES 6309/HAYASHI T/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/CATALYTIC CYCLE OF RHODIUM-CATALYZED ASYMMETRIC 1,4-ADDITION OF ORGANOBORONIC ACIDS - ARYLRHODIUM, OXA-PI-ALLYLRHODIUM, AND HYDROXORHODIUM INTERM EDI ATES/NO FIELD/KEYWORDS PLUS: ARYLBORONIC ACIDS; CONJUGATE ADDITION; ALPHA.BETA- UNSATURATED ESTERS; COUPLING REACTIONS; ENONES; REAGENTS; COMPLEXES; ALDEHYDES; ALDIMINES; LIGANDS 7743/ITOOKA R/2003/JOURNAL OF ORGANIC CHEMISTRY/RHODIUM-CATALYZED 1,4-ADDITION OF ARYLBORONIC ACIDS TO ALPHA.BETA-UNSATURATED CARBONYL-COMPOUNDS - LARGE ACCELERATING EFFECTS OF BASES AND LIGANDS/NO FIELD/KEYWORDS PLUS: ENANTIOSELECTIVE CONJUGATE ADDITION; ASYMMETRIC 1,4-ADDITION; ORGANOBORONIC ACIDS; GRIGNARD- REAGENTS; COUPLING REACTIONS; BASIC CONDITIONS; AQUEOUS- MEDIUM; CYCLIC ENONES; ALDEHYDES; COMPLEXES 8335/YOSH1DA K/2003/JOURNAL OF ORGANIC CHEM1STRY/GENERATION OF CHIRAL BORON ENOLATES BY RHODIUM-CATALYZED ASYMMETRIC 1,4-ADDITION OF 9-ARYL-9- BORABICYCLO(3.3.1)NONANES (B-AR-9BBN) TO ALPHA,BETA- UNSATURATED KETONES/NO FIELD/KEYWORDS PLUS: ARYLBORONIC ACIDS; ORGANOBORONIC ACIDS; CONJUGATE ADDITION; REAGENTS; ENONES; PHOSPHINE; LIGANDS; BINAP; CYCLOALKENONES; DERIVATIVES CLUSTER 33 2516/HAYASHI T/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/A CHIRAL CHELATING DIENE AS A NEW-TYPE OF CHIRAL LIGAND FOR TRANSITION-METAL CATALYSTS - ITS PREPARATION AND USE FOR THE RHODIUM-CATALYZED ASYMMETRIC 1.4-ADDITI0N/N0 FIELD/KEYWORDS PLUS: ARYLBORONIC ACIDS: CONJUGATE ADDITION; ALPHA.BETA- UNSATURATED KETONES; ORGANOBORONIC ACIDS; REAGENTS; ESTERS; ENONES; CYCLOALKENONES 4031/YOSH1DA K/2003/.IOURNAL OF THE AMERICAN CHEMICAL SOCIETY/A NEW CINE- SUBSTITUTION OF ALKENYL SULFONES WITH ARYLTITANIUM REAGENTS CATALYZED BY RHODIUM - MECHANISTIC STUDIES AND CATALYTIC ASYMMETRIC-SYNTHESIS OF ALLYLARENES/NO FIELD/KEYWORDS PLUS: ARYLBORONIC ACIDS: ORGANOBORONIC ACIDS; ALPHA.BETA-UNSATURATED ESTERS; CONJUGATE ADDITION; COUPLING REACTIONS; STILLE REACTION; 1,4-ADDITION; CYCLOALKENONES; COMPLEXES; LIGANDS 7237/BOITEAU JG/2003/JOURNAL OF ORGANIC CHEMISTRY/HIGH-EFFICIENCY AND ENANT1OSELECTIVITY IN 1 HE RH-CATALYZED CONJUGATE ADDITION OF ARYLBORONIC ACIDS USING MONODENTATE PHOSPHORAMIDITES/NO FIELD/KEYWORDS PLUS: TANDEM 1,4- ADDITION-ALDOL REACTION; DIFFERENTIAL ACTIVATION ENTROPY; ASYMMETRIC 1,4- ADDITION; ORGANOBORONIC ACIDS; ALPHA.BETA-UNSATURATED ESTERS; KINETIC RESOLUTION; REAGENTS; LIGANDS; ENONES; CYCLOALKENONES 191 7570/DENMARK SE/2003/JOURNAL OF ORGANIC CHEMISTRY/PALLADIUM-CATALYZED CONJUGATE ADDITION OF ORGANOSILOXANES TO ALPHA.BETA-UNSATURATED CARBONYL­ COMPOUNDS AND NITROALKENES/NO FIELD/KEYWORDS PLUS: CROSS-COUPLING REACTIONS; HYPERVALENT SILOXANE DERIVATIVES; ELECTRON-DEFICIENT OLEFINS; ASYMMETRIC 1,4- ADDITION; ARYLBORONIC ACIDS; ORGANOBORONIC ACIDS; ARYL HALIDES; PALLADIUM(O)- CATALYZED SILYLATION; BASIC CONDITIONS; SILVER(I) OXIDE 10681/SHI Q/2003/TETRAHEDRON LETTERS/BIPYRIDYL-BASED DIPHOSPHINE AS AN EFFICIENT LIGAND IN THE RHODIUM-CATALYZED ASYMMETRIC CONJUGATE ADDITION OF ARYLBORONIC ACIDS TO ALPHA,BETA-UNSATURATED KETONES/NO FIELD/KEYWORDS PLUS: CHIRAL DIPYRIDYLPHOSPHINE LIGAND; ORGANOBORONIC ACIDS; MICHAEL ADDITIONS; BETA­ KETOESTERS: 1.4- ADDITION; HYDROGENATION; CYCLOALKENONES; NITROALKENES; REAGENTS; ENONES CLUSTER 34 258/RAMACHARY DB/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/ ORGANOCATALYTIC ASYMMETRIC DOMINO KNOEVENAGEL/DIELS-ALDER REACTIONS - A BIOORGAN1C APPROACH TO THE DIASTEREOSPECIFIC AND ENANTIOSELECTIVE CONSTRUCTION OF HIGHLY SUBSTITUTED SP1RO(5,5 »UNDECANE-1,5,9-TRIONES/AUTHOR KEYWORDS: AMINO ACIDS; ASYMMETRIC CATALYSIS; CYCLOADDITION; DOMINO REACTIONS; ENANTIOSELECTIVITY/KEYWORDS PLUS: AM1NO-ACID-DERIVAT1VES; MANNICH-TYPE REACTIONS; ALPHA.BETA-UNSATURATED KETONES; ALPHA-AMINO; ORGANIC-SYNTHESIS; ALDOL REACTIONS; ALDEHYDES; CHEMISTRY; CATALYSTS; ROUTE 12760/RAMACHARY DB/2002/TETRAHEDRON LETTERS/AMINE-CATALYZED DIRECT SELF DIELS- ALDER REACTIONS OF ALPHA.BETA-UNSATURATED KETONES IN WATER - SYNTHESIS OF PRO­ CHIRAL CYCLOHEXANONES/AUTHOR KEYWORDS: AMINES; CYCLOHEXANONES; DIELS-ALDER REACTIONS; ORGANOCATALYS1S; ENAMINES; IMINES; AQUEOUS MEDIA/KEYWORDS PLUS: ASYMMETRIC ALDOL REACTIONS; PROLINE 13439/THAYUMANAVAN R/2002/TETRAHEDRON LETTERS/AMINE-CATALYZED DIRECT DIELS- ALDER REACTIONS OF ALPHA.BETA- UNSATURATED KETONES WITH NITRO OLEFINS/AUTHOR KEYWORDS: AMINES; CATALYSIS; CYCLOHEXANONES; DIELS ALDER REACTIONS; ENAMINES; MICHAEL REACTIONS/KEYWORDS PLUS: ASYMMETRIC ALDOL REACTIONS; ENANTIOSELECTIVE SYNTHESIS; MICHAEL ADDITIONS; 4-NITROCYCLOHEXANONES; 2-AMINO- 1.3-BUTADIENES; CYCLOADDITION; NITROALKENES; DERIVATIVES CLUSTER 35 714/GADEMANN K/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/THE 4TH HELICAL SECONDARY STRUCTURE OF BETA-PEPTIDES - THE (P)-2(8)-HELIX OF A BETA-HEXAPEPTIDE CONSISTING OF (2R,3S)-3- AMINO-2-HYDROXY ACID RESIDUES/AUTHOR KEYWORDS: AMINO ACIDS; BETA-PEPTIDES; CONFORMATIONAL ANALYSIS; PEPTIDOMIMETICS; SECONDARY STRUCTURE/KEYWORDS PLUS: AMINO-ACID; OLIGOMERS; FOLDAMERS; DESIGN 1552/MART1NEK TA/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/CIS-2- AMINOCYCLOPENTANECARBOXYLIC ACID OLIGOMERS ADOPT A SHEET-LIKE STRUCTURE - SWITCH FROM HELIX TO NONPOLAR STRAND/AUTHOR KEYWORDS: AMINO ACIDS; CHIRALITY; CONFORMATION ANALYSIS; NMR SPECTROSCOPY; PEPTIDES/KEYWORDS PLUS: BETA-PEPTIDES; SECONDARY STRUCTURE; SIDE- CHAINS; SPECTROSCOPY; TURNS 2121/SHARMA GVM/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/ROBUST MIXED 10/12 HELICES PROMOTED BY ALTERNATING CHIRALITY IN A NEW FAMILY OF C-LINKED CARBO- BETA-PEPT1DES/NO FIELD/KEYWORDS PLUS: SUGAR AMINO-ACIDS: SECONDARY STRUCTURE; OLIGOMERS: DESIGN; GLYCOBIOLOGY; FOLDAMERS; NMR CLUSTER 36 1332/PARK HG/2002/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/HIGHLY ENANTIOSELECTIVE AND PRACTICAL CINCHONA-DERIVED PHASE- TRANSFER CATALYSTS FOR THE SYNTHESIS OF ALPIIA-AMINO-ACIDS/NO FIELD/KEYWORDS PLUS: QUATERNARY AMMONIUM-SALTS; ASYMMETRIC- SYNTHESIS; TRANSFER ALKYLATION; DERIVATIVES; IMINES; SOLVENT 192 3686/001 T/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/DESIGN OF N-SPIRO C-2- SYMMETRIC CHIRAL QUATERNARY AMMONIUM BROMIDES AS NOVEL CHIRAL PHASE­ TRANSFER CATALYSTS - SYNTHESIS AND APPLICATION TO PRACTICAL ASYMMETRIC- SYNTHESIS OF ALPHA-AMINO-ACIDS/NO FIELD/KEYWORDS PLUS: HIGHLY ENANTIOSELECTIVE SYNTHESIS; WEITZ- SCHEFFER EPOXIDATION; L-DOPA ESTERS; MICHAEL-ADDITION; STEREOSELECTIVE SYNTHESIS; D-GLUCOSE; CINCHONA ALKALOIDS; DARZENS REACTION; AZACROWN ETHERS; CROWN-ETHERS 11318/PARK HG/2003/TETRAHEDRON LETTERS/HIGHLY EFFICIENT ORTHO-FLUORO-DIMERIC CINCHONA-DERIVED PHASE- TRANSFER CATALYSTS/NO FIELD/KEYWORDS PLUS: ALPHA- AMINO-ACIDS; ENANTIOSELECTIVE SYNTHESIS; ASYMMETRIC-SYNTHESIS 12126/SHIBUGUCH1 T/2002/TETRAHEDRON LETTERS/DEVELOPMENT OF NEW ASYMMETRIC 2- CENTER CATALYSTS IN PHASE- TRANSFER REACTIONS/AUTHOR KEYWORDS: PHASE-TRANSFER CATALYSIS; ASYMMETRIC 2- CENTER CATALYST; ALKYLATION; MICHAEL ADDITION/KEYWORDS PLUS: ALPHA-AMINO-ACIDS; ENANTIOSELECTIVE SYNTHESIS; ALKYLATION; DERIVATIVES; EPOXIDATION; IMINE CLUSTER 37 17/HILLS 1D/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/TOWARD AN IMPROVED UNDERSTANDING OF 1 TIE UNUSUAL REACTIVITY OF PD(0)/TRIALKYLPHOSPHANE CATALYSTS IN CROSS-COUPLINGS OF ALKYL ELECTROPHILES - QUANTIFYING THE FACTORS THAT DETERMINE THE RAFE OF OXIDATIVE ADDITION/AUTHOR KEYWORDS: CROSS-COUPLING; HOMOGENEOUS CATALYSIS; PALLADIUM PHOSPHANE; LIGANDS; REACTION MECHANISMS/KEYWORDS PLUS: GRIGNARD-REAGENTS; ARYLBORONIC ACIDS; SUZUKI REACTIONS; MILD CONDITIONS; BETA-HYDROGENS; EFFICIENT; CHLORIDES; HALIDES; ARYL: DERIVATIVES 125/TANG HF/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/LIGANDS FOR PALLADIUM- CATALYZED CROSS-COUPLINGS OF ALKYL- HALIDES - USE OF AN ALKYLDIAMINOPHOSPHANE EXPANDS THE SCOPE OF THE STILLE REACTION/AUTHOR KEYWORDS: CROSS-COUPLING; HOMOGENEOUS CATALYSIS; PALLADIUM; PHOSPHANE LIGANDS; STILLE REACT ION/KEYWORDS PLUS: GRIGNARD-REAGENTS; SUZUKI REACTIONS; BETA- HYDROGENS; BOND FORMATION; EFFICIENT; CHLORIDES: BROMIDES; DERIVATIVES; CENTERS; ACIDS 1936/ZHOU JR/2003/.IOURNAL OF THE AMERICAN CHEMICAL SOCIETY/CROSS-COUPL1NGS OF UNACTIVATED SECONDARY ALKYL-HALIDES - ROOM-TEMPERATURE NICKEL-CATALYZED NEGISH1 REACTIONS OF ALKYL BROMIDES AND IODIDES/NO FIELD/KEYWORDS PLUS: SUZUKI REACTIONS; BETA-HYDROGENS; EFFICIENT: TOSYLATES; CENTERS 2107/ECKHARDT M/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/THE 1ST APPLICATIONS OF CARBENE LIGANDS IN CROSS-COUPLINGS OF ALKYL ELECTROPHILES - S0N0GASH1RA REACTIONS OF UNACTIVATED ALKYL BROMIDES AND IODIDES/NO FIELD/KEYWORDS PLUS: GRIGNARD-REAGENTS; SUZUKI REACTIONS; BETA­ HYDROGENS; PALLADIUM(II); COMPLEXES; CHLORIDES; TOSYLATES; HALIDES 2376/ZHOU JR/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/PALLADIUM-CATALYZED NEG1SH1 CROSS-COUPLING REACTIONS OF UNACTIVATED ALKYL IODIDES, BROMIDES, CHLORIDES. AND TOSYLATES/NO FIELD/KEYWORDS PLUS: GRIGNARD-REAGENTS; SUZUKI REACT IONS; BETA- HYDROGENS; EFFICIENT; HALIDES; DERIVATIVES; SECONDARY; CENTERS 3542/LEE JY/2003/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/ROOM-TEMPERATURE HIYAMA CROSS-COUPLINGS OF ARYLSILANES WITH ALKYL BROMIDES AND IODIDES/NO FIELD/KEYWORDS PLUS: SUZUKI REACTIONS; CHLORIDES; BONDS CLUSTER 38 722/MAR1GO M/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/CATALYTIC, HIGHLY ENANTIOSELECTIVE, DIRECT AMINATION OF BETA- KETOESTERS/AUTHOR KEYWORDS: BETA­ KETOESTERS; AMINATION; ASYMMETRIC CATALYSIS; COPPER; SYNTHETIC METHODS/KEYWORDS PLUS: AMIN0-ACID-DERIVAT1VES; ASYMMETRIC ALPHA- AMINATION; CHIRAL ZIRCONIUM CATALYST; MANNICH-TYPE REACTIONS; STRECKER REACTION; HYDROGEN-CYANIDE; IMINO ESTERS; COMPLEXES; ALDEHYDES; ALCOHOLS 193 1577/BOGEVlG A/2002/ANGEW ANDTE CHEMIE-INTERNATIONAL EDITION/DIRECT ORGANO- CATALYT1C ASYMMETRIC ALPHA-AMINATION OF ALDEHYDES - A SIMPLE APPROACH TO OPTICALLY-ACTIVE ALPHA- AMINO ALDEHYDES, ALPHA-AMINO ALCOHOLS. AND ALPHA- AMINO-ACIDS/AUTHOR KEYWORDS: ALDEHYDES; AMINO ACIDS: AMINO ALCOHOLS; AMINO ALDEHYDES; ASYMMETRIC CATALYSIS/KEYWORDS PLUS: ENANTIOSELECTIVE SYNTHESIS; ZIRCONIUM CATALYST; STRECKER REACTION; HYDROGEN-CYANIDE; ALDOL REACTIONS; STRATEGIES; IMINES; ARYLGLYCINES; DERIVATIVES; ALKYLATION 6095/KUMARAGURUBARAN N/2002/.IOURNAL OF THE AMERICAN CHEMICAL SOCIETY/DIRECT L- PROLINE-CATALYZED ASYMMETRIC ALPHA-AMINATION OF KETONES/NO FIELD/KEYWORDS PLUS: AMINO-ACID DERIVATIVES; ORGANIC CATALYSIS; ENANTIOSELECTIVE AMINATION; ZIRCONIUM CATALYST; UNMODIFIED KETONES; STRECKER REACTION; HYDROGEN-CYANIDE; MANNICH REACTION; HYDRAZINO ACIDS; ALDOL REACTIONS CLUSTER 39 830/REETZ MT/2003/ANGEWANDTE CHEMIE-INTERNATIONAL EDITION/A NEW PRINCIPLE IN COMBINATORIAL ASYMMETRIC TRANSITION-METAL CATALYSIS - MIXTURES OF CHIRAL MONODENTATE P LIGANDS/AUTHOR KEYWORDS: ASYMMETRIC CATALYSIS; COMBINATORIAL CHEMISTRY; HYDROGENATION; PHOSPHORUS; RHODIUM/KEYWORDS PLUS: ENANTIOSELECTIVE HYDROGENATION: ENAMIDES; DISCOVERY 4674/PENA D/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/HIGHLY ENANTIOSELECT IVE RHODIUM-CATALYZED HYDROGENATION OF BETA-DEHYDROAMINO ACID- DERIVATIVES USING MONODENTATE PHOSPHORAMIDITES/NO FIELD/KEYWORDS PLUS: AMINO ACIDS; ASYMMETRIC HYDROGENA1 ION; LIGANDS 10962/NAKANO D/2003/TETRAHEDRON LET! ERS/ENANTIOSELECTIVE HYDROGENATION OF ITACONATE USING RHODIUM BIHELICENOL PHOSPHITE COMPLEX - MATCHED/MISMATCHED PHENOMENA BETWEEN HELICAL AND AXIAL CHIRALITY/NO FIELD/KEYWORDS PLUS: ASYMMETRIC HYDROGENATION; CATALYZED HYDROGENATION; LIGANDS 12520/REETZ MT/2002/TETRAHEDRON LETTERS/ENANT1OSELECTIVE HYDROGENATION OF ENAMIDES CATALYZED BY CHIRAL RHODIUM-MONODENTATE PHOSPHITE COMPLEXES/NO FIELD/KEYWORDS PLUS: ASYMMETRIC HYDROGENATION; LIGANDS CLUSTER 40 8552/BOSE DS/2003/JOURNAL OF ORGANIC CHEMISTRY/GREEN CHEMISTRY APPROACHES TO THE SYNTHESIS OF 5- ALKOXYCARBONYL-4-ARYL-3,4-DIHYDROPYRIMIDIN-2( 1 H)-ONES BY A 3- COMPONENT COUPLING OF ONE-POT CONDENSATION REACTION - COMPARISON OF ETHANOL, WATER. AND SOLVENT-FREE CONDITIONS/NO FIELD/KEYWORDS PLUS: BIGINELLI DIHYDROPYRIMIDINE SYNTHESIS; EFFICIENT SYNTHESIS; ORGANIC-SYNTHESIS; 3-COMPONENT; CATALYSIS; BLOCKERS; PROTOCOL 10315/REDDY KR/2003/TETRAHEDRON LETTERS/NEW ENVIRONMENTALLY FRIENDLY SOLVENT- FREE SYNTHESIS OF DIHYDROPYRIMIDINONES CATALYZED BY N-BUTYL-N.N-DIMETHYL- ALPHA- PHENYLETHYLAMMONIUM BROMIDE/AUTHOR KEYWORDS: N-BUTYL-N.N-DIMETHYL- ALPHA- PHENYLETHYLAMMONIUM BROMIDE; BETA-KETOESTERS; BIGINELLI REACTION; DIHYDROPYRIMIDINONES/KEYWORDS PLUS: CALCIUM-CHANNEL MODULATORS; ONE-POT SYNTHESIS: BIGINELLI REACTION; EFFICIENT SYNTHESIS; CONDENSATION REACTION; 3- COMPONENT; 3.4-DIHYDROPYRIMIDIN-2( 1 H)-ONES; REVISION; PROTOCOL; CHLORIDE 10774/TU S/2003/TETRAHEDRON LETTERS/ONE-POT SYNTHESIS OF 3.4-DIHYDROPYRIMIDIN-2(IH)- ONES USING BORIC-ACID AS CATALYST/NO FIELD/KEYWORDS PLUS: CALCIUM-CHANNEL BLOCKERS: BIGINELLI REACTION; ANTIHYPERTENSIVE AGENTS; MICROWAVE IRRADIATION; FLUOROUS SYNTHESIS; DIHYDROPYRIMIDINONES; ESTERS; CHLORIDE; 5- ALKOXYCARBONYL- 4-ARYL-3.4-DIHYDROPYRIMIDIN-211 H)-ONES; REVISION 11355/PARASKAR AS/2003/TETRAHEDRON LETTERS/CU(OTF)(2) - A REUSABLE CATALYST FOR HIGH-YIELD SYNTHESIS OF 3.4-DIHYDROPYRIMIDIN-2( 11 l)-ONES/AUTHOR KEYWORDS: ALDEHYDES; BIGINELLI REACTIONS; CATALYSTS; 3.4-DIHYDROPYRIMIDIN-2(IH)-ONES; COPPER AND COMPOUNDS/KEYWORDS PLUS: ONE-POT SYNTHESIS; BIGINELLI DIHYDROPYRIMIDINE SYNTHESIS; SOLVENT-FREE CONDITIONS; CONDENSATION REACTION; EFFICIENT SYNTHESIS: IMPROVED PROTOCOL; 3-COMPONENT: CHLORIDE 194 11458/SALEHl P/2003/TETRAHEDRON LETTERS/SILICA SULFURIC-ACID - AN EFFICIENT AND REUSABLE CATALYST FOR THE ONE-POT SYNTHESIS OF 3,4-DIHYDROPYRIMIDIN-2(lH)- ONES/AUTHOR KEYWORDS: BIGINELLI REACTION; SILICA SULFURIC ACID; DIHYDROPYRIMIDINONES; CATALYSIS; SOLID PHASE/KEYWORDS PLUS: BIGINELLI DIHYDROPYRIMIDINE SYNTHESIS; CALCIUM-CHANNEL BLOCKERS; HETEROGENEOUS SYSTEM; MILD CONDITIONS; CONDENSATION REACTION; ESTERS; 3-COMPONENT; ACID/ NAN02; PROTOCOL; MONASTROL 11520/MA1TI G/2003/TETRAHEDRON LETTERS/ONE-POT SYNTHESIS OF DIHYDROPYRIMIDINONES CATALYZED BY LITHIUM BROMIDE - AN IMPROVED PROCEDURE FOR THE BIGINELLI REACT1ON/AUTHOR KEYWORDS: DIHYDROPYRIMIDINONES; LITHIUM BROMIDE; BIGINELLI REACTION/KEYWORDS PLUS: ALDEHYDES; REVISION 11968/SHAABANI A/2003/TETRAHEDRON LETTERS/AMMONIUM CHLORIDE-CATALYZED ONE-POT SYNTHESIS OF 3,4- DIHYDROPYRIMIDIN-2-( 1H)-ONES UNDER SOLVENT-FREE CONDITIONS/AUTHOR KEYWORDS; BIGINELLI REACTION; DIHYDROPYRIMIDINONES; AMMONIUM CHLORIDE; ONE-POT CONDENSATION; SOLVENT-FREE/KEYWORDS PLUS: BIGINELLI REACTION; EFFICIENT SYNTHESIS; CONDENSATION REACTION; PARALLEL SYNTHESIS; DIHYDROPYRIMIDINONES; 3-COMPONENT; ACID; REVISION; PROTOCOL; ESTERS CLUSTER 41 10046/HUANG JW/2003/TETRAIIEDRON LETTERS/LEWIS-ACID BF3-CENTER-DOT-OET2- CATALYZED FRIEDEL-CRAFTS REACTION OF METHYLENECYCLOPROPANES WITH ARENES/AUTHOR KEYWORDS: METHYLENECYCLOPROPANES; LEWIS ACID; AROMATIC COMPOUNDS; FRIEDEL-CRAFTS REACTION; RING-OPENING REACTION/KEYWORDS PLUS: PALLADIUM; ALKYLIDENECYCLOPROPANES; (3+2)- CYCLOADDITION; PRONUCLEOPHILES 12455/SHI M/2002/TETRAHEDRON LETTERS/A NOVEL RING-OPENING REACTION OF METHYLENECYCLOPROPANES WITH AROMATIC-AMINES CATALYZED BY LEWIS- ACIDS/AUTHOR KEYWORDS: METHYLENECYCLOPROPANES (MCPS); AROMATIC AMINES; ALIPHATIC AMINES; RING-OPENING REACTIONS LEWIS ACIDS/KEYWORDS PLUS: PALLADIUM; ALKYLIDENECYCLOPROPANES; PRONUCLEOPHILES; (3+2)-CYCLOADDITION; HYDROCARBONATION; ALKENES 13726/XU B/2002/TETRAHEDRON LETTERS/THE REACTIONS OF THIOLS AND DIPHENYLDISULFIDE WITH TERMINALLY SUBSTITUTED METHYLENECYCLOPROPANES/AUTHOR KEYWORDS; THIOL; METHYLENECYCLOPROPANE; RADICAL REACTION/KEYWORDS PLUS: CATALYZED ADDITION; RADICAL CLOCKS; PALLADIUM; ACETYLENES CLUSTER 42 5649/PASCALY M/2002/.IOURNAL OF THE AMERICAN CHEMICAL SOCIETY/DNA-MEDIATED CHARGE-TRANSPORT - CHARACTERIZATION OF A DNA RADICAL LOCALIZED AT AN ARTIFICIAL NUCLEIC-ACID BASE/NO FIELD/KEYWORDS PLUS: FLASH-QUENCH TECHNIQUE; RANGE ELECTRON- TRANSFER; DUPLEX DNA; HOLE TRANSPORT; HHAL METHYLTRANSFERASE; DISTANCE DEPENDENCE; FERROCYTOCHROME-C; HOPPING MECHANISM; OXIDATIVE DAMAGE; PULSE-RADIOLYSIS 6785/WILL1AMS TT/2002/JOURNAL OF THE AMERICAN CHEMICAL SOCIETY/THE EFFECT OF VARIED ION DISTRIBUTIONS ON LONG-RANGE DNA CHARGE-TRANSPORT/NO FIELD/KEYWORDS PLUS: ELECTRON-TRANSFER; DUPLEX DNA; OXIDATION; GUANINE; ASSEMBLIES; SEQUENCE; DAMAGE; FILMS; BASES 7599/DELANEY S/2003/JOURNAL OF ORGANIC CHEMISTR Y/LONG-RANGE DNA CHARGE- TRANSPORT/NO FIELD/KEYWORDS PLUS: PHOTOINDUCED ELECTRON-TRANSFER; OXIDATIVE DAMAGE; CRYSTAL-STRUCTURE; GUANINE OXIDATION; LIGHT-SWITCH; DUPLEX DNA; DISTANCE; SEQUENCE; BASE; MECHANISM CLUSTER 43 7205/CLIVE DLJ/2003/JOURNAL OF ORGANIC CHEMISTRY/DERIVATIZED AMINO-ACIDS RELEVANT TO NATIVE PEP TIDE-SYNTHESIS BY CHEMICAL LIGATION AND ACYL TRANSFER/NO FIELD/KEYWORDS PLUS: UNPROTECTED PEPTIDES; STAUD1NGER LIGATION; PROTEIN­ SYNTHESIS; SELENOCYSTEINE; SEGMENTS; ETHERS; AUXILIARY; REMOVAL; ESTERS; AZIDE 195 10752/KA W AKAMI T/2003/TETRAHEDRON LETTERS/A PHOTOREMOVABLE LIGATION AUXILIARY FOR USE IN POLYPEPTIDE- SYNTHESIS/NO FIELD/KEYWORDS PLUS: NATIVE CHEMICAL LIGATION: PROTEINS: THIOESTER 11095/MERKX R/2003/TETRAHEDRON LETTERS/CHEMOSELECTIVE COUPLING OF PEPTIDE- FRAGMENTS USING THE STAUDINGER LIGATION/AUTHOR KEYWORDS: AMIDE-FORMING LIGATION; AZIDO PEPTIDES; PEPTIDE O-(DIPHENYLPHOSPHINE)PHENYL ESTERS; STAUDINGER LIGATION/KEYWORDS PLUS: NATIVE CHEMICAL LIGATION; DIAZO TRANSFER; PROTEINS; AUXILIARY: AZIDE; ACIDS CLUSTER 44 9712/BALAN D/2002/JOURNAL OF’ ORGANIC CHEMISTRY/TITANIUM ISOPROPOXIDE AS EFFICIENT CATALYST FOR THE AZA- BAYLIS-HILLMAN REACTION - SELECTIVE FORMATION OF ALPHA­ METHYLENE-BET A-AMINO ACID-DERIVATIVES/NO FIELD/KEYWORDS PLUS: ACTIVATED DOUBLE-BONDS; TRICARBONYLCHROMIUM COMPLEXES; CARBONYL-COMPOUNDS; ESTERS; ALDEHYDES; ROUTE 11540/BALAN D/2003/TETRAHEDRON LETTERS/CHIRAL QUINUCLIDINE-BASED AMINE CATALYSTS FOR THE ASYMMETRIC ONE-POT. 3-COMPONENT AZA-BAYLIS-HILLMAN REACTION/NO FIELD/KEYWORDS PLUS: METHYL VINYL KETONE: ACID-DERIVATIVES; TRICARBONYLCHROMIUM COMPLEXES; SELECTIVE FORMATION; VERSION; IMINES 13849/CICLOSI M/2002/TETRAH EDRON LETTERS/SYNTHESIS OF UNSATURATED BETA-AMINO ACID-DERIVATIVES FROM CARBAMATES OF THE BAYLIS-HILLMAN PRODUCTS/NO FIELD/KEYWORDS PLUS: CONVENIENT APPROACH; PYRROLIDIN-2-ONES; BASICITY; CYCLIZATION; ALDEHYDES: SUPPORT; MN(IIl): ESTERS; DBU 196 APPENDIX 5 THE COMPARISON OF TWO PARTITIONS IN CASE 2 The Dispersion of Articles over Clusters for Two Partitions. In the following table, columns A-D show the dispersion of articles in clusters generated by the field expert over the clusters generated by the complete link cluster method whereas columns E-H show the dispersion of articles in clusters generated by the complete link cluster method over the clusters generated by the field expert. A B C D E F G H Complete Complete Expert Expert Complete Complete Expert Expert Doc.nr. Clu.nr. Doc.nr. Clu.nr. Doc.nr. Clu.nr. Doc.nr. Clu.nr. 10297 4 10297 1 1014 1 1014 2 12144 4 12144 1 7888 1 7888 5 12579 18 12579 1 12254 1 12254 2 9986 19 9986 1 13307 1 13307 2 11762 20 11762 1 14376 1 14376 2 8568 24 8568 1 7415 2 7415 14 879 28 879 1 7732 2 7732 14 1143 30 1143 1 8686 2 8686 14 7743 32 7743 1 7672 3 7672 14 12760 34 12760 1 8098 3 8098 14 1936 37 1936 1 9348 3 9348 14 10774 40 10774 1 10297 4 10297 1 13849 44 13849 1 12144 4 12144 1 1014 1 1014 2 12650 4 12650 2 12254 1 12254 2 13490 4 13490 2 13307 1 13307 2 10480 4 10480 5 14376 1 14376 2 6342 5 6342 15 12650 4 12650 2 11268 5 11268 15 13490 4 13490 2 12470 5 12470 15 12402 8 12402 2 723 6 723 12 12575 8 12575 2 1112 6 1112 12 12730 8 12730 2 1579 6 1579 12 11709 10 11709 2 495 7 495 5 1397 13 1397 2 4105 7 4105 13 4996 13 4996 2 4502 7 4502 14 13502 13 13502 2 12402 8 12402 2 1508 18 1508 2 12575 8 12575 2 13761 19 13761 2 12730 8 12730 2 14370 19 14370 2 1640 9 1640 5 11271 20 11271 2 534 9 534 7 13699 20 13699 2 1631 9 1631 7 779 22 779 2 13651 9 13651 7 1844 26 1844 2 8856 9 8856 8 10348 26 10348 2 11709 10 11709 2 13195 26 13195 2 7018 10 7018 5 384 27 384 2 12554 10 12554 5 1123 28 1123 2 5691 11 5691 5 1168 28 1168 2 7916 11 7916 5 197 1539 28 1539 2 8761 11 8761 5 4813 28 4813 2 9194 11 9194 5 10703 29 10703 2 14233 11 14233 5 672 30 672 2 9617 11 9617 6 1260 30 1260 2 480 12 480 7 7570 33 7570 2 742 12 742 7 13439 34 13439 2 743 12 743 7 2107 37 2107 2 10188 12 10188 7 2376 37 2376 2 11032 12 11032 7 3542 37 3542 2 11244 12 11244 7 6095 38 6095 2 1397 13 1397 2 8552 40 8552 2 4996 13 4996 2 10315 40 10315 2 13502 13 13502 2 11520 40 11520 2 2480 13 2480 3 11968 40 11968 2 9759 13 9759 3 10046 41 10046 2 4707 14 4707 3 12455 41 12455 2 3645 14 3645 5 13726 41 13726 2 1417 14 1417 14 2480 13 2480 3 3252 14 3252 14 9759 13 9759 3 6107 14 6107 14 4707 14 4707 3 347 15 347 5 13041 18 13041 3 704 15 704 5 10156 19 10156 3 851 15 851 5 8536 24 8536 3 7933 15 7933 5 1117 26 1117 3 8726 15 8726 5 9166 26 9166 3 2509 16 2509 5 1142 27 1142 3 8657 16 8657 6 12227 27 12227 3 10449 16 10449 9 9928 29 9928 3 13672 16 13672 14 10364 29 10364 3 6608 17 6608 4 1089 30 1089 3 6005 17 6005 5 1090 31 1090 3 1763 17 1763 6 1463 31 1463 3 2373 17 2373 6 1734 31 1734 3 4114 17 4114 6 11514 31 11514 3 5370 17 5370 6 5305 32 5305 3 6288 17 6288 6 10681 33 10681 3 12579 18 12579 1 11318 36 11318 3 1508 18 1508 2 12126 36 12126 3 13041 18 13041 3 125 37 125 3 1509 18 1509 7 11355 40 11355 3 11164 18 11164 7 11458 40 11458 3 12141 18 12141 I 9712 44 9712 3 12714 18 12714 7 6608 17 6608 4 13407 18 13407 7 8335 32 8335 4 9986 19 9986 1 3686 36 3686 4 13761 19 13761 2 7888 1 7888 5 14370 19 14370 2 10480 4 10480 5 10156 19 10156 3 495 7 495 5 11762 20 11762 1 1640 9 1640 5 11271 20 11271 2 7018 10 7018 5 13699 20 13699 2 198 12554 10 12554 5 12125 21 12125 5 5691 11 5691 5 1376 21 1376 6 7916 11 7916 5 10175 21 10175 6 8761 11 8761 5 779 22 779 2 9194 11 9194 5 4113 22 4113 5 14233 11 14233 5 3605 22 3605 6 3645 14 3645 5 2207 23 2207 17 347 15 347 5 2381 23 2381 17 704 15 704 5 3086 23 3086 17 851 15 851 5 3910 23 3910 17 7933 15 7933 5 6823 23 6823 17 8726 15 8726 5 8568 24 8568 1 2509 16 2509 5 8536 24 8536 3 6005 17 6005 5 3357 24 3357 10 12125 21 12125 5 1703 25 1703 15 4113 22 4113 5 2025 25 2025 15 5089 32 5089 5 3739 25 3739 15 7237 33 7237 5 4779 25 4779 15 258 34 258 5 1844 26 1844 2 722 38 722 5 10348 26 10348 2 1577 38 1577 5 13195 26 13195 2 4674 39 4674 5 1117 26 1117 3 10962 39 10962 5 9166 26 9166 3 12520 39 12520 5 384 27 384 2 9617 11 9617 6 1142 27 1142 3 8657 16 8657 6 12227 27 12227 3 1763 17 1763 6 879 28 879 1 2373 17 2373 6 1123 28 1123 2 4114 17 4114 6 1168 28 1168 2 5370 17 5370 6 1539 28 1539 2 6288 17 6288 6 4813 28 4813 2 1376 21 1376 6 10703 29 10703 2 10175 21 10175 6 9928 29 9928 3 3605 22 3605 6 10364 29 10364 3 2516 33 2516 6 1143 30 1143 1 1332 36 1332 6 672 30 672 2 830 39 830 6 1260 30 1260 2 11540 44 11540 6 1089 30 1089 3 534 9 534 7 1090 31 1090 3 1631 9 1631 7 1463 31 1463 3 13651 9 13651 7 1734 31 1734 3 480 12 480 7 11514 31 11514 3 742 12 742 7 7743 32 7743 1 743 12 743 7 5305 32 5305 3 10188 12 10188 7 8335 32 8335 4 11032 12 11032 7 5089 32 5089 5 11244 12 11244 7 6309 32 6309 13 1509 18 1509 7 7570 33 7570 2 11164 18 11164 7 10681 33 10681 3 12141 18 12141 7 7237 33 7237 5 12714 18 12714 7 2516 33 2516 6 199 13407 18 13407 7 4031 33 4031 13 8856 9 8856 8 12760 34 12760 1 10449 16 10449 9 13439 34 13439 2 3357 24 3357 10 258 34 258 5 7205 43 7205 11 714 35 714 16 10752 43 10752 11 1552 35 1552 16 11095 43 11095 11 2121 35 2121 16 723 6 723 12 11318 36 11318 3 1112 6 1112 12 12126 36 12126 3 1579 6 1579 12 3686 36 3686 4 4105 7 4105 13 1332 36 1332 6 6309 32 6309 13 1936 37 1936 1 4031 33 4031 13 2107 37 2107 2 7415 2 7415 14 2376 37 2376 2 7732 2 7732 14 3542 37 3542 2 8686 2 8686 14 125 37 125 3 7672 3 7672 14 17 37 17 14 8098 3 8098 14 6095 38 6095 2 9348 3 9348 14 722 38 722 5 4502 7 4502 14 1577 38 1577 5 1417 14 1417 14 4674 39 4674 5 3252 14 3252 14 10962 39 10962 5 6107 14 6107 14 12520 39 12520 5 13672 16 13672 14 830 39 830 6 17 37 17 14 10774 40 10774 1 6342 5 6342 15 8552 40 8552 2 11268 5 11268 15 10315 40 10315 2 12470 5 12470 15 11520 40 11520 2 1703 25 1703 15 11968 40 11968 2 2025 25 2025 15 11355 40 11355 3 3739 25 3739 15 11458 40 11458 3 4779 25 4779 15 10046 41 10046 2 5649 42 5649 15 12455 41 12455 2 6785 42 6785 15 13726 41 13726 2 7599 42 7599 15 5649 42 5649 15 714 35 714 16 6785 42 6785 15 1552 35 1552 16 7599 42 7599 15 2121 35 2121 16 7205 43 7205 11 2207 23 2207 17 10752 43 10752 11 2381 23 2381 17 11095 43 11095 11 3086 23 3086 17 13849 44 13849 1 3910 23 3910 17 9712 44 9712 3 6823 23 6823 17 11540 44 11540 6 200 APPENDIX 6 BIBLIOGRAPHIC DESCRIPTIONS OF CORE DOCUMENT CLUSTERS IN CASE 2. The bibliographic descriptions are delimited to document number and title. CLUSTER 1 1703 DYNAMICS AND ENERGETICS OF HOLE TRAPPING IN DNA BY 7- DEAZAGUAN1NE 2746 DIRECT OBSERVATION OF GUANINE RADICAL-CAI ION DEPROTONATION IN DUPLEX DNA USING PULSE-RADIOLYSIS 3394 RAPID RADICAL FORMATION BY DNA CHARGE-TRANSPORT THROUGH SEQUENCES LACKING INTERVENING GUANINES 3498 BASE SEQUENCE EFFECTS IN RADICAL-CATION MIGRATION IN DUPLEX DNA - SUPPORT FOR THE POLARON-LIKE HOPPING MODEL 3676 RATIONAL DESIGN OF A DNA WIRE POSSESSING AN EXTREMELY HIGH HOLE TRANSPORT ABILITY 3739 DYNAMICS AND ENERGETICS OF SINGLE-STEP HOLE TRANSPORT IN DNA HAIRPINS 5965 N-2-PHENYLDEOXYGUANOSINE - MODULATION OF THE CHEMICAL- PROPERTIES OF DEOXYGUANOSINE TOWARD ONE-ELECTRON OXIDATION IN DNA 6342 DYNAMICS OF INTERSTRAND AND INTRASTRAND HOLE TRANSPORT IN DNA HAIRPINS CLUSTER! 17 TOWARD AN IMPROVED UNDERSTANDING OF THE UNUSUAL REACTIVITY OF PD(0)/TR1ALKYLPHOSPHANE CATALYSTS IN CROSS-COUPLINGS OF ALKYL ELECTROPHILES- QUANTIFYING THE FACTORS THAT' DETERMINE THE RATE OF OXIDATIVE ADDITION 125 LIGANDS FOR PALLADIUM-CATALYZED CROSS-COUPLINGS OF ALKYL- HALIDES - USE OF AN ALKYLDIAMINOPHOSPHANE EXPANDS THE SCOPE OF THE STILLE REACTION 1936 CROSS-COUPLINGS OF UNACTIVATED SECONDARY ALKYL-HALIDES-ROOM- TEMPERATURE NICKEL-CATALYZED NEGISHI REACTIONS OF ALKYL BROMIDES AND IODIDES 2376 PALLADIUM-CATALYZED NEGISHI CROSS-COUPLING REACTIONS OF UNACTIVATED ALKYLIODIDES. BROMIDES. CHLORIDES, AND TOSYLATES 4813 BORONIC ACIDS - NEW COUPLING PARTNERS IN ROOM-TEMPERATURE SUZUKI REACTIONS OF ALKYL BROMIDES - CRYSTALLOGRAPHIC CHARACTERIZATION OF AN OXIDATIVE-ADDITION ADDUCT GENERATED UNDER REMARKABLY MILD CONDITIONS CLUSTER 3 1089 A PRACTICAL AND HIGHLY-ACTIVE RUTHENIUM-BASED CATALYST THAT EFFECTS THE CROSS METATHESIS OF ACRYLONITRILE 1142 A SELF-GENERATING. HIGHLY-ACTIVE. AND RECYCLABLE OLERIN- METATHESIS CATALYST 1463 ANEW HIGHLY EFFICIENT RUTHENIUM METATHESIS CATALYST 1734 A HIGHLY-ACTIVE AND AIR-STABLE RUTHENIUM COMPLEX FOR OLEFIN METATHESIS 201 12579 SYNTHESIS OF A HIGHLY FUNCTIONALIZED TRICYCLIC RING-SYSTEM RELATED TO GUANACASTEPENE VIA A TANDEM RING-CLOSING METATHESIS REACTION CLUSTER 4 1014 CATALYTIC. ASYMMETRIC BAYLIS-HILLMAN REACTION OF IMINES WITH METHYL VINYLKETONE AND METHYL ACRYLATE 9712 TITANIUM 1SOPROPOXIDE AS EFFICIENT CATALYST FOR THE AZA- BAYLIS-HILLMAN REACTION - SELECTIVE FORMATION OF ALPHA- METHYLENE-BETA-AMINO ACID- DERIVATIVES 12254 ONE-POT AZA-BAYLIS-HILLMAN REACTIONS OF ARYLALDEHYDES AND DIPHENYLPHOSPHINAMIDE WITH METHYL VINYL KETONE IN THE PRESENCE OF TICL4PPH3. AND ET3N 13307 BAYLIS-HILLMAN REACTIONS OF N- ARYLIDENEDIPHENYLPHOSPHINAMIDES WITH METHYL VINYL KETONE. METHYL ACRYLATE. AND ACRYLONITRILE 14376 LEWIS BASE AND L-PROLINE CO-CATALYZED BAYLIS-HILLMAN REACTION OF ARYLALDEHYDES WITH METHYL VINYL KETONE CLUSTERS 11355 CU(OTF)(2) - A REUSABLE CATALYST FOR HIGH-YIELD SYNTHESIS OF 3.4- DIHYDROPYRIMIDIN-2(1 H)-ONES 11458 SILICA SULFURIC-ACID - AN EFFICIENT AND REUSABLE CATALYST FOR THE ONE-POT SYNTHESIS OF 3.4-DIHYDROPYRIMIDIN-2( IH»-ONES CLUSTER 6 2516 A CHIRAL CHELATING DIENE AS A NEW-TYPE OF CHIRAL LIGAND FOR TRANSITION­ METAL CATALYSIS - ITS PREPARATION AND USE FOR THE RHODIUM-CATALYZED ASYMMETRIC 1,4-ADDITION 4031 A NEW CINE-SUBSTITUTION OF ALKENYL SULFONES WI TH ARYLTITANIUM REAGENTS CATALYZED BY RHODIUM - MECHANISTIC STUDIES AND CATALYTIC ASYMMETRIC- SYNTHESIS OF ALLYLARENES 5089 RHODIUM-CATALYZED ASYMMETRIC 1,4-ADDITION OF ARYLTITANIUM REAGENTS GENERATING CHIRAL TITANIUM ENOLATES - ISOLATION AS SILYL ENOL ETHERS 5305 A NEW-TYPE OF CATALYTIC TANDEM 1.4-ADDITION-ALDOL REACTION WHICH PROCEEDS THROUGH AN (OXA-PI-ALLYL)RI IODIUM INTERMEDIATE 6309 CATALYTIC CYCLE OF RHODIUM-CAI'ALYZED ASYMMETRIC 1,4-ADDITION OF ORGANOBORON1C ACIDS - ARYLRHODIUM, OXA-PI-ALLYLRHODIUM, AND HYDROXORHODIUM INTERMEDIATES 7743 RHODIUM-CATALYZED 1,4-ADDITION OF ARYLBORONIC ACIDS TO ALPHA.BETA- UNSATURATED CARBONYL-COMPOUNDS - LARGE ACCELERATING EFFECTS OF BASES AND LIGANDS 8335 GENERATION OF CHIRAL BORON ENOLATES BY RHODIUM-CATALYZED ASYMMETRIC 1,4-ADDITION OF 9-ARYL-9- BORABICYCLO(3.3.1)NONANES (B-AR-9BBN) TO ALPHA.BETA- NSATURATED KETONES 10681 BIPYRIDYL-BASED DIPHOSPHINE AS AN EFFICIENT LIGAND IN THE RHODIUM- CATALYZED ASYMMETRIC CONJUGATE ADDITION OF ARYLBORONIC ACIDS TO ALPHA,BETA-UNSATURATED KETONES 202 262 347 704 779 851 3605 4113 4502 6095 7933 11273 12554 13439 Cluster 7 A FACILE AND RAPID ROUTE TO HIGHLY ENANTIOPURE 1,2-DIOLS BY NOVEL CATALYTIC ASYMMETRIC ALPHA-AM1NOXYLATION OF ALDEHYDES THE DIRECT AND ENANTIOSELECTIVE. ONE-POT, 3-COMPONENT, CROSS- MANNICH REACTION OF ALDEHYDES THE 1ST ORGANOCATALYT1C ENANTIOSELECTIVE INVERSE-ELECTRON- DEMAND HETERO-DIELS-ALDER REACTION PROLINE-CATALYZED ASYMMETRIC ALPHA-AMINATION OF ALDEHYDES AND KETONES - AN ASTONISHINGLY SIMPLE ACCESS TO OPTICALLY- ACTIVE ALPHA- HYDRAZINOCARBONYL-COMPOUNDS HIGHLY ENANTIOSELECTIVE ORGANOCATALY TIC CONJUGATE ADDITION OF MALONATES TO ACYCLIC ALPHA,BETA-UNSATURATED ENONES NOVEL SMALL ORGANIC-MOLECULES FOR A HIGHLY ENANTIOSELECTIVE DIRECT ALDOL REACTION DIRECT CATALYTIC ASYMMETRIC MICHAEL REACTION OF HYDROXYKETONES - ASYMMETRIC ZN CATALYSIS WITH A ET2ZN/L1NKED- BINOL COMPLEX KINETIC AND STEREOCHEMICAL EVIDENCE FOR THE INVOLVEMENT OF ONLY ONE PROLINE MOLECULE IN THE TRANSITION-STATES OF PROLINE-CATALYZED INTRAMOLECULAR AND INTERMOLECULAR ALDOL REACTIONS DIRECT L-PROLINE-CATALYZED ASYMMETRIC ALPHA-AMINATION OF KETONES DIRECT ENANTIOSELECTIVE MICHAEL ADDITION OF ALDEHYDES TO VINYL KETONES CATALYZED BY CHIRAL AMINES PROLINE CATALYZED ALDOL REACTIONS IN AQUEOUS MICELLES - AN ENVIRONMENTALLY FRIENDLY REACTION SYSTEM ANTI-SELECTIVE SMP-CATALYZED DIRECT ASYMMETRIC MANNICH-TYPE REACTIONS - SYNTHESIS OF FUNCTIONALIZED AMINO-ACID DERIVATIVES AMINE-CATALYZED DIRECT DIELS-ALDER REAC TIONS OF ALPHA,BETA- UNSATURATED KETONES WITH NITRO OLEFINS 203 APPENDIX 7 BIBLIOGRAPHIC DESCRIPTIONS OF CLUSTERS WITH A SIZE > 3 IN CASE 3 Bibliographic data as follows: record number/ first author name/ publication year/ Journal name/ title/ author key words/key words plus. Missing data is indicated by “No Field”. CLUSTER 1 711/PATYI 1/2003/MATHEMATISCHE ANNALEN/ON THE OKA PRINCIPLE IN A BANACH-SPACE. I/NO FIELD/NO FIELD 712/PATYI I/2003/MATHEMA1 ISCHE ANNALEN/ON THE OKA PRINCIPLE IN A BANACH-SPACE. Il/NO FIELD/NO FIELD 475/LEITERER J/2003/.IOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/A RELATIVE OKA-GRAUERT PRINCIPLE ON 1-CONVEX SPACES/NO FIELD/NO FIELD CLUSTER! 205/KARPENKO N/2003/1NVENTIONES MATHEMAI ICAE/ESSENTIAL DIMENSION OF QUADRICS/NO FIELD/KEYWORDS PLUS: FIELDS 207/KARPENKO NA/2003/INVENTIONES MATHEMATICAE/ON THE 1ST Will' INDEX OF QUADRATIC-FORMS/NO FIELD/NO FIELD 463/MERKURJEV A/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/STEENROD OPERATIONS AND DEGREE FORMULAS/NO FIELD/NO FIELD CLUSTER 3 454/GRIGORYAN A/2002/JOURNAL DE MATHEMATIQUES PURES ET APPLIQUEES/HITTING PROBABILITIES FOR BROWNIAN-MO LION ON RIEMANN1AN- MANIFOLDS/NO FIELD/KEYWORDS PLUS: HEAT KERNEL; MAXIMUM PRINCIPLE; DIFFUSION; BOUNDS 686/MURATA M/2003/MATHEMAT1SCHE ANNALEN/HEAT ESCAPE/NO FIELD/KEYWORDS PLUS: POSITIVE CAUCHY-PROBLEM; PARABOLIC HARNACK INEQUALITY; LOCAL DIRICHLET SPACES; INTRINSIC METRIC APPROACH; RIEMANNIAN-MANIFOLDS; ELLIPTIC-OPERATORS; SEMISMALL PERTURBA LIONS; FUNDAMENTAL-SOLUTIONS; MARTIN BOUNDARIES; EQUATIONS 788/GRIGORYAN A/2002/MATHEMAT1SCHE ANNALEN/HARNACK INEQUALITIES AND SUB­ GAUSSIAN ESTIMATES FOR RANDOM- WALKS/NO FIELD/KEYWORDS PLUS: BROWNIAN-MOTION: RIEMANNIAN-MANIFOLDS; S1ERPINSKI CARPET; HEAT KERNEL: GRAPHS; NASH CLUSTER 4 215/LEHN M/2003/INVENTIONES MATHEMATICAE/THE CUP PRODUCT OF HILBERT SCHEMES FOR K3 SURFACES/NO FIELD/KEYWORDS PLUS: COHOMOLOGY RING; POINTS; ALGEBRA; SHEAVES 559/LI WP/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/STABILITY OF THE COHOMOLOGY RINGS OF HILBERT SCHEMES OF POINTS ON SURFACES/NO FIELD/NO FIELD 665/LI WP/2002/.IOURNAL FUR DIE REINE UND ANGEWANDTE: MATHEMATIK/ON BLOWUP FORMULAS FOR THE S-DUALITY CONJECTURE OF VAFA AND WITTEN III - RELATIONS WITH VERTEX OPERATOR-ALGEBRAS/NO FIELD/KEYWORDS PLUS: SURFACES: MONSTER; NUMBERS; SHEAVES; POINTS 803/LI WP/2002/MATHEMATISCHE ANNALEN/VERTEX ALGEBRAS AND THE COHOMOLOGY RING STRUCTURE OF HILBERT SCHEMES OF POINTS ON SURFACES/NO FIELD/KEYWORDS PLUS; VARIETY; PRODUCE 204 CLUSTERS 240/LEARY IJ/2003/INVENTIONES MATHEMATICAE/SOME GROUPS OF TYPE VF/NO FIELD/KEYWORDS PLUS: FINITENESS PROPERTIES; ARTIN GROUPS; K-THEORY; CLASSIFYING SPACE; DISCRETE-GROUPS; ASSEMBLY MAPS; GRAPH GROUPS; FP-INFINITY; SUBGROUPS; CHARACTERS 279/LUCK W/2002/INVENTIONES MATHEMATICAE/THE RELATION BETWEEN THE BAUM-CONNES CONJECT URE AND THE TRACE CONJECTURE/NO FIELD/KEYWORDS PLUS: COUNTEREXAMPLE 656/LUCK W/2002/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/CHERN CHARACTERS FOR PROPER EQUIVARIANT HOMOLOGY THEORIES AND APPLICATIONS TO K- THEORY AND L-THEORY/NO FIELD/KEYWORDS PLUS: BAUM-CONNES CONJECTURE: DISCRETE- GROUPS; PROOF CLUSTER 6 61/RAMAKRISHNA R/2002/ANNALS OF MATHEMATICS/DEFORMING GALOIS REPRESENTATIONS AND THE CONJECTURES OF SERRE AND FONTAINE-MAZUR/NO FIELD/KEYWORDS PLUS: MODULAR-REPRESENTATIONS; ELLIPTIC-CURVES; CONSTRUCTION; DEFORMATION; THEOREM; FORMS 192/KHARE C/2003/INVENTIONES MATHEMATICAE/FINITENESS OF SELMER GROUPS AND DEFORMATION RINGS/NO FIELD/KEYWORDS PLUS: GALOIS REPRESENTATIONS; EVEN REPRESENTATION 193/KHARE C/2003/INVENT10NES MATHEMATICAE/ON ISOMORPHISMS BETWEEN DEFORMATION RINGS AND HECKE RINGS/NO FIELD/KEYWORDS PLUS: MODULAR-REPRESENTATIONS; ELLIPTIC- CURVES; FORMS CLUSTER 7 173/WANG BX/2002/COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS/THE LIMIT BEHAVIOR OF SOLUTIONS FOR THE CAUCHY-PROBLEM OF THE COMPLEX GINZBURG-LANDAU EQUATION/NO FIELD/KEYWORDS PLUS: NONLINEAR SCHRODINGER-EQUATIONS; DISPERSIVE EQUATIONS; KLEIN-GORDON; H-S; SCATTERING 798/MASMOUDI N/2002/MATHEMATISCHE ANNALEN/FROM NONLINEAR KLEIN-GORDON EQUATION TO A SYSTEM OF COUPLED NONLINEAR SCHRODINGER-EQUATIONS/NO FIELD/KEYWORDS PLUS: GLOBAL CAUCHY-PROBLEM; ENERGY SCATTERING 857/MACHIHARA S/2002/MATHEMATISCHE ANNALEN/NONRELATIVISTIC LIMIT IN THE ENERGY SPACE FOR NONLINEAR KLEIN-GORDON EQUATIONS/NO FIELD/KEYWORDS PLUS: GLOBAL CAUCHY-PROBLEM; SCHRODINGER-EQUATION CLUSTERS 220/KOBAYASHI S/2003/INVENTIONES MATHEMATICAE/IWASAWA THEORY FOR ELLIPTIC- CURVES AT SUPERSINGULAR PRIMES/NO FIELD/KEYWORDS PLUS: ABELIAN-VARIETIES; RATIONAL-POINTS; REDUCTION; FIELDS; TOWERS 281/KURIHARA M/2002/INVENTIONES MATHEMATICAE/ON THE TATE SHAFAREVICH GROUPS OVER CYCLOTOMIC FIELDS OF AN ELLIPTIC CURVE WITH SUPERSINGULAR REDUCTION I/NO FIELD/KEYWORDS PLUS: P-ADIC REPRESENTATIONS; IWASAWA THEORY; RATIONAL-POINTS; FORMAL GROUPS; CONJECTURE; BIRCH 499/KURIHARA M/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/IWASAWA THEORY AND FITTING IDEALS/NO FIELD/KEYWORDS PLUS; TOTALLY-REAL FIELDS; MINUS CLASS-GROUPS; ELLIPTIC-CURVES; ABELIAN-FIELDS; NUMBER-FIELDS; SELMER GROUPS; CONJECTURE; EXTENSIONS; RESIDUE; FORMULA 205 CLUSTER 9 437/REY O/2002/JOURNAL DE MATHEMATIQUES PURES ET APPLIQUEES/THE QUESTION OF INTERIOR BLOW-UP POINTS FOR AN ELLIPTIC NEUMANN PROBLEM - THE CRITICAL CASE/NO FIELD/KEYWORDS PLUS: CRITICAL SOBOLEV EXPONENTS: MULTI-PEAKED SOLUTIONS; LEAST­ ENERGY SOLUTIONS; CAHN-HILLIARD EQUATION; CRITICAL NONLINEARITY; STATIONARY SOLUTIONS: EXISTENCE; BEHAVIOR; EQUILIBRIA; SYMMETRY 630/GUI CF/2002/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/ESTIMATES FOR BOUNDARY-BUBBLING SOLUTIONS TO AN ELLIPTIC NEUMANN PROBLEM/NO FIELD/KEYWORDS PLUS: CRITICAL SOBOLEV EXPONEN TS; CRITICAL NONLINEARITY; EQUATIONS; BEHAVIOR 738/GROSSI M/2003/MATHEMATISCHE ANNALEN/A UNIQUENESS RESULT FOR A NEUMANN PROBLEM INVOLVING THE CRITICAL SOBOLEV EXPONENT/NO FIELD/NO FIELD CLUSTER 10 99/LI AB/2003/COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS/ON SOME CONFORMALLY INVARIANT FULLY NONLINEAR EQUATIONS/NO FIELD/KEYWORDS PLUS: MONGE-AMPERE TYPE; ELLIPTIC-EQUATIONS; DIRICHLET PROBLEM; SCALAR CURVATURE; GEOMETRY; 4-MANIFOLDS; MANIFOLDS; SYMMETRY 494/SCHWETLICK H/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/CONVERGENCE OF THE YAMABE FLOW FOR LARGE ENERGIES/NO FIELD/KEYWORDS PLUS: CONFORMALLY FLAT MANIFOLDS; SCALAR CURVATURE; COMPACTNESS 535/GUAN PF/2003/.IOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/A FULLY NONLINEAR CONFORMAL FLOW ON LOCALLY CONFORMALLY FLAT MANIFOLDS/NO FIELD/KEYWORDS PLUS: POSITIVE RICCI CURVATURE: SCALAR CURVATURE; YAMABE FLOW; GEOMETRY; EQUATIONS CLUSTER 11 269/GROVE K/2002/INVENTIONES MATHEMATICAE/COHOMOGENEITY ONE MANIFOLDS WITH POSITIVE RICCI CURVATURE/NO FIELD/KEYWORDS PLUS: HOMOGENEOUS EINSTEIN-METRICS; SPHERES; SPACES 295/WILKING B/2002/INVENT10NES MATHEMATICAE/MANIFOLDS WITH POSITIVE SECTIONAL CURVATURE ALMOST EVERYWHERE/NO FIELD/NO FIELD 334/BELEGRADEK I/2003/.IOURNAL OF THE AMERICAN MATHEMATICAL SOCIETY/OBSTRUCTIONS TO NONNEGATIVE CURVATURE AND RATIONAL HOMOTOPY- THEORY/AUTHOR KEYWORDS: NONNEGATIVE CURVATURE; SOUL; DERIVATION; HALPERINS CONJECTURE/KEYWORDS PLUS: STRICTLY POSITIVE CURVATURE; HOMOGENEOUS SPACES; VECTOR-BUNDLES; MANIFOLDS; FINITENESS: SOUL CLUSTER 12 84/FORNI G/2002/ANNALS OF MATHEMATICS/DEVIATION OF ERGODIC AVERAGES FOR AREA­ PRESERVING FLOWS ON SURFACES OF HIGHER-GENUS/NO FIELD/KEYWORDS PLUS: INTERVAL EXCHANGE TRANSFORMATIONS; QUADRATIC- DIFFERENTIALS; MEASURED FOLIATIONS; TEICHMULLER-SPACES: DYNAMICAL-SYSTEMS; EXPONENTS: BILLIARDS; MANIFOLD 200/KONTSEV1CH M/2003/INVENT10NES MATHEMATICAE/CONNECTED COMPONENTS OF THE MODULI SPACES OF ABELIAN DIFFERENTIALS WITH PRESCRIBED SINGULARITIES/NO FIELD/KEYWORDS PLUS: QUADRATIC-DIFFERENTIALS; TRANSFORMATIONS; FOLIATIONS 859/MUCINORAYMUNDO J/2002/MA THEMATISCHE ANNALEN/COMPLEX STRUCTURES ADAPTED TO SMOOTH VECTOR-FIELDS/NO FIELD/KEYWORDS PLUS: QUADRATIC-DIFFERENTIALS; PRESCRIBED SINGULARITIES; FLOWS CLUSTER 13 242/ESNAULT H/2003/1NVENTIONES MATHEMATICAE/VARIET1ES OVER A FINITE-FIELD WITH TRIVIAL CHOW GROUP OF 0- CYCLES HAVE A RATIONAL POINT/NO FIELD/KEYWORDS PLUS: RIGID COHOMOLOGY; HODGE TYPE 206 628/CHlARELLOTTO B/2002/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/A COMPARISON THEOREM FOR WEIGHTS/NO FIELD/KEYWORDS PLUS: UNIPOTENT F-ISOCRYSTALS; RIGID COHOMOLOGY: PURITY 869/BESSER A/2002/MATHEMATISCHE ANNALEN/COLEMAN INTEGRATION USING THE TANNAK1AN FORMALISM/NO FIELD/KEYWORDS PLUS: UNIPOTENT F-ISOCRYSTALS; RIGID COHOMOLOGY CLUSTER 14 363/EMERTON M/2002/.IOURNAL OF THE AMERICAN MATHEMATICAL SOC1ETY/SUPERSINGULAR ELLIPTIC-CURVES, THETA-SERIES AND WEIGHT 2 MODULAR-FORMS/NO FIELD/KEYWORDS PLUS: HECKE OPERATORS: MOD-P; REPRESENTATIONS 667/HALBERSTADT E/2002/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/FERMAT- CURVES - RESULTS AND PROBLEMS/NO FIELD/KEYWORDS PLUS: ELLIPTIC-CURVES; MODULAR- REPRESENTATIONS; LAST THEOREM; PRIME 677/EMERTON M/2003/MATHEMATISCHE ANNALEN/OPTIMAL QUOTIENTS OF MODULAR JACOBIANS/NO FIELD/KEYWORDS PLUS: ELLIPTIC-CURVES; HECKE OPERATORS; REPRESENTATIONS; FORMS; PARAMETRIZATIONS CLUSTER 15 635/BERTIN MA/2002/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/ON THE REGULARITY OF VARIETIES HAVING AN EX TREMAL SECANT LINE/NO FIELD/KEYWORDS PLUS: CASTELNUOVO 804/GIRALDO L/2002/MATHEMATISCHE ANNALEN/ON THE PROJECTIVE-NORMALITY OF ENRIQUES SURFACES (WITH AN APPENDIX BY LOPEZ.ANGELO,FELICE AND VERRA,ALESSANDRO)/NO FIELD/KEYWORDS PLUS: VECTOR-BUNDLES; ALGEBRAIC-SURFACES; KOSZUL COHOMOLOGY; LINEAR-SYSTEMS; CURVE; SYZYGIES; VARIETIES; DIMENSION 871/NOMA A/2002/MATHEMATISCHE ANNALEN/A BOUND ON THE CASTELNUOVO-MUMFORD REGULARITY FOR CURVES/NO FIELD/KEYWORDS PLUS: SPACE CLUSTER 16 261/FANG FQ/2002/INVENTIONES MATHEMATICAE/THE 2ND TWISTED BETTI NUMBER AND THE CONVERGENCE OF COLLAPSING RIEMANNIAN-MANIFOLDS/NO FIELD/KEYWORDS PLUS: POSITIVE SECTIONAL CURVATURE; DIAMETER; FINITENESS; GEOMETRY; HOMOTOPY 486/ANDERSON MT/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/SCALAR CURVATURE AND THE EXISTENCE OF GEOMETRIC STRUCTURES ON 3-MANIFOLDS, II/NO FIELD/KEYWORDS PLUS: COLLAPSING RIEMANNIAN-MANIFOLDS; VACUUM EINSTEIN EQUATIONS; METRIC DEGENERATIONS 564/ANDERSON MT/2002/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/SCALAR CURVATURE AND THE EXISTENCE OF GEOMETRIC STRUCTURES ON 3-MANIFOLDS, I/NO FIELD/KEYWORDS PLUS: COLLAPSING RIEMANNIAN-MANIFOLDS; VACUUM EINSTEIN EQUATIONS; METRIC DEGENERATIONS 867/TUSCHMANN W/2002/MATHEMATISCHE ANNALEN/GEOMETRIC D1FFEOMORPHISM FINITENESS IN LOW DIMENSIONS AND HOMOTOPY GROUP FINITENESS/NO FIELD/KEYWORDS PLUS: COLLAPSING RIEMANNIAN-MANIFOLDS; CONTROLLED TOPOLOGY; BOUNDING HOMOTOPY; CURVATURE; THEOREMS; DIAMETER CLUSTER 17 12/KASPAROV G/2003/ANNALS OF MATHEMATICS/GROUPS ACTING PROPERLY ON BOLIC SPACES AND THE NOVIKOV- CONJECTURE/NO FIELD/KEYWORDS PLUS: BAUM-CONNES CONJECTURE; EQUIVARIANT KK-THEORY; ALGEBRAS 312/LAFFORGUE V/2002/INVENTIONES MATHEMAT1CAE/BIVARIANT K-THEORY FOR BANACH- ALGEBRAS AND BAUM-CONNES CONJECTURE/NO FIELD/KEYWORDS PLUS: CROSSED-PRODUCTS; FREES: PROOF; AMENABILITY; PROPERTY 207 472/EMERSON H/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/NONCOMMUTATIVE POINCARE-DUALITY FOR BOUNDARY ACTIONS OF HYPERBOLIC GROUPS/NO FIELD/KEYWORDS PLUS: NOVIKOV-CONJECTURE; KK-THEORY; ALGEBRAS CLUSTER 18 18/JIANG DH/2003/ANNALS OF MATHEMAT1CS/THE LOCAL CONVERSE THEOREM FOR SO(2N+1) AND APPLICATIONS/NO FIELD/KEYWORDS PLUS: P-ADIC FIELD; RANKIN-SELBERG CONVOLUTIONS; MULTIPLICITY ONE THEOREM; GENERIC REPRESENTATIONS; THETA- LIFT; GL(N); CONJECTURE: MODULES; PROOF 21/LAPID E/2003/ANNALS OF MATHEMATICS/ON THE NONNEGATIVITY OF L(l/2. PI) FOR SO2N+1/NO FIELD/KEYWORDS PLUS: AUTOMORPHIC L-FUNCTIONS; IRREDUCIBLE REPRESENTATIONS; PLANCHEREL MEASURES; ADIC GROUPS: REDUCTIBILITY; GL(N); CLASSIFICATION; CONJECTURE: VALUES; FIELD 72/K.IM HH/2002/ANNALS OF MATHEMATICS/FUNCTORIAL PRODUCTS FOR GL(2)XGL(3) AND 'HIE SYMMETRIC CUBE FOR GL(2)/NO FIELD/KEYWORDS PLUS: RANKIN-SELBERG CONVOLUTIONS: LANGLANDS- SHAHIDI METHOD; AUTOMORPHIC L-FUNCTIONS; CUSP FORMS; INTERTWINING- OPERATORS; FOURIER COEFFICIENTS; PLANCHEREL MEASURES; REPRESENTATIONS; CONJECTURE; GL(N) 73/BUSHNELL CJ/2002/ANNALS OF MATHEMATICS/APPENDIX - ON CERTAIN DYADIC REPRESENTATIONS/NO FIELD/KEYWORDS PLUS; RANKIN-SELBERG CONVOLUTIONS; GL(N) 349/KIM HH/2003/JOURNAL OF THE AMERICAN MATHEMATICAL SOCIETY/FUNCTORIALITY FOR THE EXTERIOR SQUARE OF GL(4) AND THE SYMMETRIC 4TH OF GL(2)/NO FIELD/KEYWORDS PLUS: P-ADIC FIELD: LOCAL LANGLANDS CONJECTURE; AUTOMORPHIC L-FUNCTIONS; 1NTERTWIN1NG-OPERATORS; PLANCHEREL MEASURES; EULER PRODUCTS; SHAHIDI METHOD: CUSP FORMS; GL(N); REPRESENTATIONS 364/MOEGLIN C/2002/JOURNAL OF THE AMERICAN MATHEMATICAL SOCIETY/CONSTRUCTION OF DISCRETE-SERIES FOR CLASSICAL P-ADIC GROUPS/AUTHOR KEYWORDS: CLASSICAL GROUPS; P- ADIC FIELDS; IRREDUCIBLE SQUARE INTEGRABLE REPRESENTATIONS; IRREDUCIBLE TEMPERED REPRESENTATIONS; NONUNITARY DUAL; LOCAL LANGLANDS CORRESPONDENCES/KEYWORDS PLUS: INDUCED REPRESENTATIONS; INTERTW1NING- OPERATORS; PLANCHEREL MEASURES; REDUCIBILITY; GL(N); NORMALIZATION; CONJECTURE; INDUCTION; PROOF CLUSTER 19 523/11UYBRECHTS D/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/ FINITENESS RESULTS FOR COMPACT HYPERKAHLER MANIFOLDS/NO FIELD/KEYWORDS PLUS: KAHLER-MANIFOLDS 642/MARKMAN E/2002/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/GENERATORS OF THE COHOMOLOGY RING OF MODULI SPACES OF SHEAVES ON SYMPLECTIC SURFACES/NO FIELD/KEYWORDS PLUS: PROJECTIVE VARIETY; HILBERT SCHEME; EQUATIONS; MANIFOLDS 716/HUYBRECHTS D/2003/MATHEMAT1SCHE ANNALEN/THE KAHLER CONE OF A COMPACT HYPERKAHLER MANIFOLD/NO FIELD/NO FIELD 779/NAMIKAWA Y/2002/MATHEMAT1SCHE ANNALEN/COUNTER-EXAMPLE TO GLOBAL TORELLI PROBLEM FOR IRREDUCIBLE SYMPLECTIC-MAN1FOLDS/NO FIELD/NO FIELD CLUSTER 20 276/JONSSON M/2002/1NVENT10NES MATHEMATICAE/STABLE MANIFOLDS OF HOLOMORPHIC DIFFEOMORPHISMS/NO FIELD/NO FIELD 426/D1NH TC/2003/JOURNAL DE MATHEMATIQUES PURES ET APPLIQUEES/DYNAMICS OF POLYNOMIAL-LIKE MAPPINGS/NO FIELD/KEYWORDS PLUS: ENTROPY; CURRENTS; DIFFEOMORPHISMS; EXPONENTS; MAP 875/DINH TC/2002/MATHEMATISCHE ANNALEN/PERMUTABLE HOLOMORPHIC ENDOMORPHISMS OF P-K/NO FIELD/KEYWORDS PLUS: DYNAMICS 208 CLUSTER 21 94/KOHN RV/2003/COMMUNICAT1ONS ON PURE AND APPLIED MATHEMATICS/UPPER BOUND ON THE COARSENING RATE FOR AN EPITAXIAL-GROWTH MODEL/NO FIELD/KEYWORDS PLUS: MOLECULAR-BEAM EPITAXY; SLOPE SELECTION; PHASE-TRANSITIONS; THIN-FILMS; CONTINUUM MODEL; GRADIENT THEORY; DYNAMICS; ENERGY; COMPACTNESS; DESORPTION 124/AMBROSIO L/2003/COMMUN1CATIONS ON PURE AND APPLIED MATHEMATICS/A VISCOSITY PROPERTY OF MINIMIZING MICROMAGNETIC CONFIGURATIONS/NO FIELD/KEYWORDS PLUS: ENERGY; COMPACTNESS; FIELDS 150/DESIMONE A/2002/COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS/A REDUCED THEORY FOR THIN-FILM MICROMAGNETICS/NO FIELD/KEYWORDS PLUS: ENERGY; FERROMAGNETISM; COMPACTNESS 162/CONTI S/2002/COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS/A GAMMA­ CONVERGENCE RESULT FOR THE 2-GRAD1ENT THEORY OF PHASE-TRANSIT! ON S/NO FIELD/KEYWORDS PLUS: NONCONVEX VARIATIONAL-PROBLEMS; MINIMAL INTERFACE CRITERION; SINGULAR PERTURBATIONS; LOCAL MINIMIZERS; ENERGY; FERROELAST1CS; MIXTURES; FIELDS 379/DELELLIS C/2003/JOURNAL DE MATHEMATIQUES PURES ET APPLIQUEES/THE RECTIFIABILITY OF ENTROPY MEASURES IN ONE SPACE DIMENSION/AUTHOR KEYWORDS: CONSERVATION LAWS; ENTROPY SOLUTIONS; SHOCKS; CONCENTRATION/KEYWORDS PLUS: ENERGY; MICROMAGNETICS: COMPACTNESS; REGULARITY CLUSTER 22 8/CHEUNG Y/2003/ANNALS OF MATHEMATICS/HAUSDORFF DIMENSION OF THE SET OF NONERGODIC DIRECTIONS/NO FIELD/KEYWORDS PLUS: FOLIATIONS 319/MCMULLEN CT/2003/JOURNAL OF THE AMERICAN MATHEMATICAL SOCIETY/BILLIARDS AND TEICHMULLER CURVES ON HILBERT MODULAR SURFACES/NO FIELD/KEYWORDS PLUS: ARITHMETIC FUCHSIAN-GROUPS; QUADRATIC- DIFFERENTIALS; TRIANGULAR BILLIARDS; EMBEDDINGS; SPACES 580/MINSKY Y/2002/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/ NONDIVERGENCE OF HOROCYCLIC FLOWS ON MODULI SPACE/NO FIELD/KEYWORDS PLUS: INTERVAL EXCHANGE TRANSFORMATIONS; HOMOGENEOUS SPACES; TEICHMULLER SPACE; FOLIATIONS; MANIFOLDS; SURFACES; MAPS; SET CLUSTER 23 309/BERNDTSSON B/2002/INVENTIONES MATHEMATICAE/THE PARTIAL-DERIVATIVE-EQUATION ON A POSITIVE CURRENT/NO FIELD/KEYWORDS PLUS: THEOREM 693/CHEN BY/2003/MATHEMATISCHE ANNALEN/THE BERGMAN METRIC ON COMPLETE KAHLER- MANIFOLDS/NO FIELD/KEYWORDS PLUS: PSEUDOCONVEX DOMAINS; THEOREM; KERNEL 830/MCNEAL JD/2002/MATHEMATISCHE ANNALEN/L-2 HARMONIC FORMS ON SOME COMPLETE KAHLER-MANIFOLDS/NO FIELD/KEYWORDS PLUS: PSEUDOCONVEX DOMAINS; BERGMAN- KERNEL; CONVEX DOMAINS; COHOMOLOGY; METRICS CLUSTER 24 15/BRENDLE S/2003/ANNALS OF MATHEMATICS/GLOBAL EXISTENCE AND CONVERGENCE FOR A HIGHER-ORDER FLOW IN CONFORMAL GEOMETRY/NO FIELD/KEYWORDS PLUS: ZETA-FUNCTION DETERMINANTS; RICCI FLOW; 4- MANIFOLDS; CURVATURE; INVARIANT; EQUATION; METRICS; EXTREMALS; SURFACES 69/CHANG SYA/2002/ANNALS OF MATHEMAT1CS/AN EQUATION OF MONGE-AMPERE TYPE IN CONFORMAL GEOMETRY, AND 4-MANIFOLDS OF POSITIVE RICCI CURVATURE/NO FIELD/KEYWORDS PLUS: 2ND-ORDER ELLIPTIC-EQUATIONS; COMPACT RIEMANN1AN- MANIFOLDS; ZETA-FUNCTION DETERMINANTS; DIRICHLET PROBLEM; CRITICAL EXPONENT; SCALAR CURVATURE; YAMABE FLOW; 4- MANIFOLDS; INEQUALITY; REGULARITY 209 108/CHANG SYA/2003/COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS/THE INEQUALITY OF MOSER AND TRUDINGER AND APPLICATIONS TO CONFORMAL GEOMETRY/NO FIELD/KEYWORDS PLUS: KAHLER-EINSTEIN METRICS; PRESCRIBING GAUSSIAN CURVATURE; ZETA-FUNCTIONAL DETERMINANTS; SIMONS HIGGS-MODEL; SCALAR CURVATURE; SOBOLEV INEQUALITIES; RIEMANNIAN-MANIFOLDS; EXTREMAL METRICS; EXISTENCE: 4-MANIFOLDS 787/BRENDLE S/2002/MATHEMAT1SCHE ANNALEN/CURV ATURE FLOWS ON SURFACES WITH BOUNDARY/NO FIELD/KEYWORDS PLUS: RICCI CLUSTER 25 154/GUZZETT1 D/2002/COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS/THE ELLIPTIC REPRESENTATION OF THE GENERAL PAINLEVE-VI EQUATION/NO FIELD/KEYWORDS PLUS: ORDINARY DIFFERENTIAL-EQUATIONS; 2- DIMENSIONAL ISING-MODEL; RATIONAL COEFFICIENTS; QUANTUM COHOMOLOGY; MONODROMY; DEFORMATION; TC 549/HERTLING C/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/TT-ASTERISK GEOMETRY. FROBENIUS MANIFOLDS. THEIR CONNECTIONS. AND THE CONSTRUCTION FOR SINGULAR1TIES/NO FIELD/KEYWORDS PLUS: MIXED HODGE-STRUCTURES; KAI ILER-MAN I FOLDS; PERIOD; MONODROMY 713/CAO HD/2003/MATHEMATISCHE ANNALEN/ON QUASI-ISOMORPHIC DGBV ALGEBRAS/NO FIELD/KEYWORDS PLUS: FROBENIUS MANIFOLD STRUCTURE; COHOMOLOGY; GRAVITY; SPACE CLUSTER 26 212/POPA M/2003/INVENTIONES MATHEMATICAE/STABLE MAPS AND QUOT SCHEMES/NO FIELD/KEYWORDS PLUS: VECTOR-BUNDLES; FLAG VARIETIES; MODULI; CURVES; SURFACES 243/G1VENTAL A/2003/INVENTIONES MATHEMATICAE/QUANTUM K-THEORY ON FLAG MANIFOLDS. FINITE-DIFFERENCE TODA- LATTICES AND QUANTUM GROUPS/NO FIELD/KEYWORDS PLUS: COHOMOLOGY 321/BUCH AS/2003/.IOURNAL OF THE AMERICAN MATHEMATICAL SOCIETY/GROMOV-WITTEN INVARIANTS ON GRASSMANNIANS/AUTHOR KEYWORDS: GROMOV-WITTEN INVARIANTS: GRASSMANNIANS; FLAG VARIETIES; SCHUBERT VARIETIES; QUANTUM COHOMOLOGY; LITTLEWOOD-RICHARDSON RULE/KEYWORDS PLUS: SCHUBERT POLYNOMIALS; FLAG MANIFOLDS; FUSION RULES; FORMULA 338/RIETSCH K/2003/JOURNAL OF THE AMERICAN MATHEMATICAL SOCIETY/TOTALLY POSITIVE TOEPLITZ MATRICES AND QUAN TUM COHOMOLOGY OF PARTIAL FLAG VARIETIES/AUTHOR KEYWORDS: FLAG VARIETIES; QUANTUM COHOMOLOGY; TOTAL POSITIVITY/KEYWORDS PLUS: SCHUBERT POLYNOMIALS; PARAMETRIZATIONS; MANIFOLDS; RINGS CLUSTER 27 79/GRODAL J/2002/ANNALS OF MATHEMATICS/HIGHER LIMITS VIA SUBGROUP COMPLEXES/NO FIELD/KEYWORDS PLUS: SPORADIC SIMPLE-GROUPS; COMPACT LIE-GROUPS; CLASSIFYING­ SPACES; FINITE-GROUPS; MODULAR-REPRESEN1 Al IONS; MACKEY FUNCTORS; P-GROUP; HOMOTOPY; COHOMOLOGY; IIOMOLOGY 230/BROTO C/2003/IN VENTION ES MATHEMATICAE/HOMOTOPY-EQUIVALENCES OF P- COMPLETED CLASSIFYING-SPACES OF FINITE-GROUPS/NO F1ELD/NO FIELD 318/BROTO C/2003/JOURNAL OF THE AMERICAN MATHEMATICAL SOC1ETY/THE HOMOTOPY­ THEORY OF FUSION SYSTEMS/AUTHOR KEYWORDS: CLASSIFYING SPACE; P-COMPLETION; FINITE GROUPS; FUSION/KEYWORDS PLUS: CLASSIFYING-SPACES; HIGHER LIMITS; DECOMPOSITION; EXTENSIONS; SUBGROUPS: DIAGRAMS; CATEGORY; MODULES; MAPS; RING CLUSTER 28 11/MOSHER L/2003/ANNALS OF MATHEMATICS/QUASI-ACTIONS ON TREES I - BOUNDED VALENCE/NO FIELD/KEYWORDS PLUS: BAUMSLAG-SOLITAR GROUPS; ISOMETRIC RIGIDITY; GEOMETRY 210 264/BONK M/2002/1NVENTIONES MATHEMATICAE/QUASI-SYMMETRIC PARAMETRIZATIONS OF 2- DIMENSIONAL METRIC SPHERES/NO FIELD/KEYWORDS PLUS: CIRCLE PACKINGS: GOOD PARAMETERIZATIONS; HYPERBOLIC BUILDINGS: MEASURE-SPACES: UNIFORMIZARON; CONSTANT; THEOREM: MAPS 571/BOURDON M/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/LP- COHOMOLOGY AND BESOV-SPACES/NO FIELD/KEYWORDS PLUS: METRIC MEASURE-SPACES; HYPERBOLIC BUILDINGS; BOUNDARY CLUSTER 29 48/SAPIR MV/2002/ANN ALS OF MATHEMATICS/1SOPERIMETRIC AND ISODIAMETRIC FUNCTIONS OF GROUPS/NO FIELD/KEYWORDS PLUS: WORD PROBLEM; INEQUALITIES 49/BIRGET JC/2002/ANNALS OF MATHEMATICS/ISOPERIMETRIC FUNCTIONS OF GROUPS AND COMPUTATIONAL,- COMPLEXITY OF THE WORD PROBLEM/NO FIELD/KEYWORDS PLUS: INEQUALITIES 425/GRIMALDI R/2003/JOURNAL DE MATHEMATIQUES PURES ET APPLIQUEES/FILLING AND SURFACES OF REVOLUTION/NO FIELD/KEYWORDS PLUS: ISOPERIMETRIC INEQUALITY; CURVATURE CLUSTER 30 33/KOZLOVSK1 OS/2003/ANNALS OF MATHEMATICS/AXIOM-A MAPS ARE DENSE IN THE SPACE OF UNIMODAL MAPS IN THE C-K TOPOLOGY/NO FIELD/KEYWORDS PLUS: DYNAMICS; POLYNOMIALS; MAPPINGS; SET 58/LYUBICH M/2002/ANNALS OF MATHEMATICS/ALMOST EVERY REAL QUADRATIC MAP IS EITHER REGULAR OR STOCHASTIC/NO FIELD/KEYWORDS PLUS: NON-LINEAR TRANSFORMATIONS; ONE-DIMENSIONAL MAPS; S-UNIMODAL MAPS; DYNAMICS; POLYNOMIALS; RENORMALIZATION: UNIVERSALITY; ITERATIONS; ATTRACTORS; FAMILIES 182/AVILA A/2003/INVENTIONES MATHEMATICAE/REGULAR OR STOCHASTIC DYNAMICS IN REAL ANALYTIC FAMILIES OF UNIMODAL MAPS/NO FIELD/KEYWORDS PLUS: ONE­ DIMENSIONAL DYNAMICS; QUADRATIC POLYNOMIALS; HOLOMORPHIC MOTIONS; INVARIANT­ MEASURES; HYPERBOLICITY; INTERVAL; BOUNDS CLUSTER 31 274/TERASOMA T/2002/INVENTIONES MATHEMAT1CAE/MIXED TATE MOTIVES AND MULTIPLE ZETA VALUES/NO FIELD/NO FIELD 298/ELBAZVINCENT P/2002/INVENTIONES MATHEMATICAE/MILNOR K-THEORY OF RINGS. HIGHER CHOW GROUPS AND APPLICATIONS/NO FIELD/KEYWORDS PLUS: ALGEBRAIC CYCLES; HOMOLOGY 790/KAHN B/2002/MATHEMATISCHE ANNALEN/THE GEISSER-L,EVINE METHOD REVISITED AND ALGEBRAIC CYCLES OVER A FINITE-FIELD/NO FIELD/KEYWORDS PLUS: BLOCH-KATO CONJECTURE; MILNOR K-THEORY; CHARACTERISTIC-P; COHOMOLOGY CLUSTER 32 117/LUO WZ/2003/COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS/MASS EQUIDISTRIBUTION FOR HECKE EIGENFORMS/NO FIELD/KEYWORDS PLUS: AUTOMORPHIC L- FUNCTIONS; COEFFICIENTS; FORMS 267/DUKE W/2002/INVENTIONES MATHEMATICAE/THE SUBCONVEXITY PROBLEM FOR ARTIN L- FUNCTIONS/NO FIELD/KEYWORDS PLUS: AUTOMORPHIC L-FUNCT1ONS; FOURIER COEFFICIENTS; MODULAR-FORMS; WEIGHT; BOUNDS: SUMS 725/HARCOS G/2003/MATHEMATISCHE ANNALEN/AN ADDITIVE PROBLEM IN THE FOURIER COEFFICIENTS OF CUSP FORMS/NO FIELD/KEYWORDS PLUS: HALF-INTEGRAL WEIGHT; SELBERG L-FUNCTIONS; DIVISOR PROBLEM; MODULAR-FORMS; SUMS; REPRESENTATIONS; OPERATORS 211 CLUSTER 33 304/BRIDGELAND T/2002/INVENT10NES MATHEMA I ICAE/FLOPS AND DERIVED CATEGOR1ES/NO FIELD/NO FIELD 507/NAMIKAWA Y/2003/JOURNAL FÜR DIE REINE UND ANGEWANDTE- MATHEMATIK/MUKAI FLOPS AND DERIVED CATEGORIES/NO FIELD/NO FIELD 647/CALDARARU A/2002/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/DERIVED CATEGORIES OF TWISTED SHEAVES ON ELLIPTIC THREEFOLDS/NO FIELD/KEYWORDS PLUS: FOURIER-MUKAI TRANSFORMS CLUSTER 34 314/BERGER L/2002/INVENTIONES MATHEMATICAE/P-ADIC REPRESENTATION AND DIFFERENTIAL-EQUATIONS/NO FIELD/KEYWORDS PLUS: CRYSTALLINE REPRESENTATIONS: IWASAWA THEORY; INDEX THEOREM; F-ISOCRYSTALS; LOCAL-FIELD; COHOMOLOGY; EXTENSIONS; CURVE 315/ANDRE Y/2002/INVENTIONES MATHEMATICAE/HASSE-ARF FILTRATIONS AND P-ADIC MONODROMY/NO FIELD/KEYWORDS PLUS: F-ISOCRYSTALS; DIFFERENTIAL-EQUATIONS; INDEX THEOREM: GALOIS THEORY; REPRESENTATIONS; CURVE 316/MEBKHOUT Z/2002/INVENTIONES MATHEMATICAE/P-ADIC ANALOG OF THE TURRITTIN THEOREM AND THE THEOREM OF P- ADIC MONODROMY/NO FIELD/KEYWORDS PLUS: DIFFERENTIAL-EQUATIONS: INDEX THEOREM; F- ISOCRYSTALS; REPRESENTATIONS; COHOMOLOGY: OPERATORS: MODULES CLUSTER 35 90/MARTEL Y/2002/ANNALS OF MATHEMAT1CS/STABILITY OF BLOW-UP PROFILE AND LOWER BOUNDS FOR BLOW-UP RATE FOR THE CRITICAL GENERALIZED KDV EQUAT1ON/NO FIELD/KEYWORDS PLUS: DE-VRIES EQUATION; KORTEWEG-DEVRIES EQUATION 333/COLLIANDER J/2003/JOURNAL OF THE AMERICAN MATHEMATICAL SOCIETY/SHARP GLOBAL WELL-POSEDNESS FOR KDV AND MODIFIED KDV ON R AND T/AUTHOR KEYWORDS: KORTEWEG- DE VRIES EQUATION; NONLINEAR DISPERSIVE EQUATIONS; BILINEAR ESTIMATES; MULTILINEAR HARMONIC ANALYSIS/KEYWORDS PLUS: KORTEWEG-DEVRIES EQUATION: SEMILINEAR WAVE- EQUATIONS; ILL-POSEDNESS; DISPERSIVE EQUATIONS: EXISTENCE: SYSTEMS; TIME; L-2 361/MARTEL Y/2002/.IOURNAL OF T HE AMERICAN MATHEMATICAL SOCIETY/BLOW-UP IN FINITE­ TIME AND DYNAMICS OF BLOW-UP SOLUTIONS FOR THE L-2-CRITICAL GENERALIZED KDV EQUATION/AUTHOR KEYWORDS: CRITICAL KDV EQUATION; FINITE TIME BLOW UP; BLOW UP RATE/KEYWORDS PLUS: KORTEWEG-DEVRIES EQUATION; DE-VRIES EQUATION CLUSTER 36 416/MALCHIODI A/2002/JOURNAL DE MATHEMATIQUES PURES ET APPLIQUEES/A PERTURBATION RESULT FOR THE WEBSTER SCALAR CURVATURE PROBLEM ON THE CR SPHERE/AUTHOR KEYWORDS: WEBSTER CURVATURE; PERTURBATION METHODS; SUBELLIPTIC EQUATIONS/KEYWORDS PLUS: ITEISENBERG-GROUP; SEMILINEAR EQUATIONS; VARIATIONAL APPROACH; YAMABE PROBLEM; HOMOCLINICS; LAPLACIAN: EXISTENCE; COMPLEX 759/DANCER EN/2003/MATHEMATISCHE ANNALEN/REAL ANALYTICITY AND NON- DEGENERACY/NO FIELD/KEYWORDS PLUS: POSITIVE SOLUTIONS; NONLINEAR EQUATIONS; ELLIPTIC-EQUATIONS; SCALAR CURVATURE; DOMAIN SHAPE; UNIQUENESS; NUMBER; BIFURCATION; WAVES 845/AMBROSETTT A/2002/MATHEMATISCHE ANNALEN/ON THE YAMABE PROBLEM AND THE SCALAR CURVATURE PROBLEMS UNDER BOUNDARY-CONDITIONS/NO FIELD/KEYWORDS PLUS: S-N; MEAN-CURVATURE; PERT URBATION; MANIFOLDS 212 CLUSTER 37 479/NAKAMAYE M/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/SESHADRI CONSTANTS AND THE GEOME TRY OF SURFACES/NO F1ELD/NO FIELD 756/HWANG JM/2003/MATHEMATISCHE ANNALEN/SESHADRI-EXCEPTIONAL FOL1ATIONS/NO FIELD/KEYWORDS PLUS: CONSTANTS: SURFACES 808/OGUISO K./2002/MATHEMATISCHE ANNALEN/SESHADRI CONSTANTS IN A FAMILY OF SURFACES/NO FIELD/KEYWORDS PLUS: VARIETIES CLUSTER 38 186/WLODARCZYK J/2003/IN VENTIONES MATHEMATICAE/TOROIDAL VARIETIES AND THE WEAK FACTORIZATION THEOREM/NO FIELD/KEYWORDS PLUS: GEOMETRIC INVARIANT-THEORY; BIRATIONAL MAPS; BLOWING-UP; SINGULARITIES; SURFACES; CHARACTERISTIC-O; THREEFOLDS; MORPHISMS; BUNDLES; RINGS 337/KAWAKITA M/2003/JOURNAL OF THE AMERICAN MATHEMATICAL SOCIETY/GENERAL ELEPHANTS OF 3-FOLD DIVISORIAL CONTRACT1ONS/AUTHOR KEYWORDS: GENERAL ELEPHANT; DIVISORIAL CONTRACTION/KEYWORDS PLUS: 3-DIMENSIONAL TERMINAL SINGULARITIES 358/ABRAMOV1CH D/2002/JOURNAL OF THE AMERICAN MATHEMATICAL SOCTETY/TORIFICATION AND FACTORIZATION OF BIRATIONAL MAPS/NO FIELD/KEYWORDS PLUS: GEOMETRIC INVARIANT-THEORY; BLOWING-UP; SINGULARITIES; SURFACES; CHARACTERISTIC-O; RESOLUTION; THREEFOLDS; MORPHISMS; VARIETIES; BUNDLES CLUSTER 39 307/ETINGOF P/2002/INVENTIONES MATHEMAT1CAE/SYMPLECTIC REFLECTION ALGEBRAS, CALOGERO-MOSER SPACE, AND DEFORMED HARISH-CHANDRA HOMOMORPHISM/NO FIELD/KEYWORDS PLUS: QUANTUM INTEGRABLE SYSTEMS; AFFINE HECKE ALGEBRAS; KAC- MOODY ALGEBRAS; DUALIZING COMPLEXES; QUIVER VARIETIES; WEYL ALGEBRA: LIE­ ALGEBRAS; OPERATORS; THEOREM; RINGS 672/NAKAJIMA H/2003/MATHEMATISCHE ANNALEN/REFLECTION FUNCTORS FOR QUIVER VARIETIES AND WEYL GROUP- ACTIONS/NO FIELD/KEYWORDS PLUS: ALE GRAVITATIONAL INSTANTONS; HYPER-KAHLER QUOTIENTS; KAC-MOODY ALGEBRAS; SPACES: SINGULARITIES; CONNECTIONS; MODULI 764/CRAWLEYBOEVEY W/2003/MATHEMAT1SCHE ANNALEN/NORMALITY OF MARSDEN- WEINSTEIN REDUCTIONS FOR REPRESENTATIONS OF QUIVERS/NO FIELD/KEYWORDS PLUS: KLEINIAN SINGULARITIES: ALE SPACES; ALGEBRAS; DEFORMATIONS; GEOMETRY CLUSTER 40 66/DEBACKER S/2002/ANNALS OF MATHEMATICS/PARAMETRIZING NILPOTENT ORBITS VIA BRUHAT-TITS THEORY/NO FIELD/KEYWORDS PLUS: MINIMAL K-TYPES: P-ADIC GROUPS 251/MOEGLIN C/2003/IN VENTIONES MATHEMATICAE/STABLE PACKS OF TEMPERED REPRESENTATIONS AND OF UNIPOTENT REDUCTION FOR SO(2N+1)/NO FIELD/KEYWORDS PLUS: CHARACTER SHEAVES; COHOMOLOGY 556/MURNAGHAN F/2003/JOURNAL FUR DIE REINE UND ANGEWANDTE MATHEMATIK/LOCAL CHARACTER EXPANSIONS OF ADMISSIBLE REPRESENTATIONS OF P-ADIC GENERAL LINEAR- GROUPS/NO FIELD/KEYWORDS PLUS: MINIMAL K-TYPES; FOURIER-TRANSFORM; SHALIKA GERMS; FIELD; GL(N); GLN 213 APPENDIX 8 THE COMPARISON OF TWO PARTITIONS IN CASE 3 The Dispersion of Articles over Clusters for Two Partitions. In the following table, columns A-D show the dispersion of articles in clusters generated by the field expert over the clusters generated by the complete link cluster method whereas columns E-H show the dispersion of articles in clusters generated by the complete link cluster method over the clusters generated by the field expert. Cluster number “0” indicates articles not clustered by the field expert on grounds of insufficent information in the bibliographic descriptions. A B C D E F G H Complete Complete Expert Expert Complete Complete Expert Expert Doc. nr. Clu.nr. Doc.nr. Clu.nr. Doc. nr. Clu.nr. Doc.nr. Clu.nr. 274 31 274 0 475 1 475 2 337 38 337 0 711 1 711 34 628 13 628 0 712 1 712 34 779 19 779 1 205 2 205 12 475 1 475 2 207 2 207 12 267 32 267 2 463 2 463 16 230 27 230 3 454 3 454 10 318 27 318 3 686 3 686 36 298 31 298 3 788 3 788 4 788 3 788 4 215 4 215 16 186 38 186 5 559 4 559 15 358 38 358 5 665 4 665 30 859 12 859 6 803 4 803 15 523 19 523 6 240 5 240 18 716 19 716 6 279 5 279 15 693 23 693 6 656 5 656 15 830 23 830 6 61 6 61 31 549 25 549 6 192 6 192 31 94 21 94 7 193 6 193 19 312 17 312 8 173 7 173 25 472 17 472 8 798 7 798 25 309 23 309 9 857 7 857 25 454 3 454 10 220 8 220 19 261 16 261 10 281 8 281 19 867 16 867 10 499 8 499 19 200 12 200 11 437 9 437 26 8 22 8 11 630 9 630 26 205 2 205 12 738 9 738 26 207 2 207 12 99 10 99 26 58 30 58 12 494 10 494 24 182 30 182 12 535 10 535 24 869 13 869 13 269 11 269 24 72 18 72 13 295 11 295 24 304 33 304 13 334 11 334 24 507 33 507 13 84 12 84 16 713 25 713 14 200 12 200 11 214 559 4 559 15 859 12 859 6 803 4 803 15 242 13 242 15 279 5 279 15 628 13 628 0 656 5 656 15 869 13 869 13 242 13 242 15 363 14 363 19 571 28 571 15 667 14 667 19 790 31 790 15 677 14 677 19 463 2 463 16 635 15 635 27 215 4 215 16 804 15 804 16 84 12 84 16 871 15 871 16 804 15 804 16 261 16 261 10 871 15 871 16 486 16 486 24 319 22 319 16 564 16 564 24 725 32 725 16 867 16 867 10 212 26 212 17 12 17 12 24 243 26 243 17 312 17 312 8 321 26 321 17 472 17 472 8 338 26 338 17 18 18 18 20 240 5 240 18 21 18 21 31 79 27 79 18 72 18 72 13 11 28 11 18 73 18 73 20 193 6 193 19 349 18 349 20 220 8 220 19 364 18 364 20 281 8 281 19 523 19 523 6 499 8 499 19 642 19 642 33 363 14 363 19 716 19 716 6 667 14 667 19 779 19 779 1 677 14 677 19 276 20 276 29 18 18 18 20 426 20 426 32 73 18 73 20 875 20 875 29 349 18 349 20 94 21 94 7 364 18 364 20 124 21 124 27 315 34 315 20 150 21 150 27 316 34 316 20 162 21 162 27 556 40 556 20 379 21 379 32 264 28 264 21 8 22 8 11 333 35 333 22 319 22 319 16 361 35 361 22 580 22 580 28 307 39 307 23 309 23 309 9 672 39 672 23 693 23 693 6 494 10 494 24 830 23 830 6 535 10 535 24 15 24 15 24 269 11 269 24 69 24 69 26 295 11 295 24 108 24 108 24 334 11 334 24 787 24 787 24 486 16 486 24 154 25 154 31 564 16 564 24 549 25 549 6 12 17 12 24 713 25 713 14 15 24 15 24 212 26 212 17 108 24 108 24 243 26 243 17 787 24 787 24 321 26 321 17 215 416 36 416 24 338 26 338 17 845 36 845 24 79 27 79 18 479 37 479 24 230 27 230 3 756 37 756 24 318 27 318 3 808 37 808 24 11 28 11 18 90 35 90 25 264 28 264 21 173 7 173 25 571 28 571 15 798 7 798 25 48 29 48 27 857 7 857 25 49 29 49 27 437 9 437 26 425 29 425 27 630 9 630 26 33 30 33 32 738 9 738 26 58 30 58 12 99 10 99 26 182 30 182 12 69 24 69 26 274 31 274 0 759 36 759 26 298 31 298 3 635 15 635 27 790 31 790 15 124 21 124 27 117 32 117 29 150 21 150 27 267 32 267 2 162 21 162 27 725 32 725 16 48 29 48 27 304 33 304 13 49 29 49 27 507 33 507 13 425 29 425 27 647 33 647 33 580 22 580 28 314 34 314 31 276 20 276 29 315 34 315 20 875 20 875 29 316 34 316 20 117 32 117 29 90 35 90 25 665 4 665 30 333 35 333 22 61 6 61 31 361 35 361 22 192 6 192 31 416 36 416 24 21 18 21 31 759 36 759 26 154 25 154 31 845 36 845 24 314 34 314 31 479 37 479 24 764 39 764 31 756 37 756 24 251 40 251 31 808 37 808 24 426 20 426 32 186 38 186 5 379 21 379 32 337 38 337 0 33 30 33 32 358 38 358 5 642 19 642 33 307 39 307 23 647 33 647 33 672 39 672 23 711 1 711 34 764 39 764 31 712 1 712 34 66 40 66 35 66 40 66 35 251 40 251 31 686 3 686 36 556 40 556 20 216 Publikationer ur serien Skrifter frAn VALFRID Enmark, Romulo: Defining the Library's Activities (ISBN 91-971457-1-X) International Publications: 1 Biblioteksstudier. Folkbibliotek i flervetenskaplig belysning. Red. Romulo Enmark. (ISBN 91-971457-0-X) Skriftserien ; 1 Biblioteken och framtiden, del I. Red. Romulo Enmark. (ISBN 91-971457-1-8) Skriftserien ; 2 Biblioteken och framtiden, del IL Red. Lars Seldén. (ISBN 91-971457-2-6) Skriftserien ; 3 Hjorland, Birger: Emnerepræsentation og informationssogning. (ISBN 91-971457-3-4) Slut Hjorland, Birger: Emnerepræsentation og informationssogning. 2. uppl. med register. (ISBN 91-971457-4-2) Skriftserien ; 4 Biblioteken, kulturen och den sociala intelligensen. Red. Lars Höglund. (ISBN 91-971457-5-0) Skriftserien ; 5 Hjorland, Birger: Faglitteratur. Kvalitet, vurdering og selektion. (ISBN 91-971457-6-9) Skriftserien ; 6 Limberg, Louise: Skolbiblioteksmodeller. Utvärdering av ett utvecklingsprojekt i Örebro län. (ISBN 91-971457-7-7) Slut Hjorland, Birger: Faglitteratur. Kvalitet, vurdering og selektion. 2. rev.udgave. (ISBN 91-971457-8-5) Skriftserien ; 8 Pettersson, Rune: Verbo-visual Communication - Presentation of Clear Messages for Information and Learning. (ISBN 91-971457-9-3) Skriftserien ; 9 Pettersson, Rune: Verbo-visual Communication - 12 Selected Papers (ISBN 91 -973090-0-1 ) Skriftserien ; 10 Limberg, Louise: Skolbiblioteksmodeller. Utvärdering av ett utvecklingsprojekt i Örebro län. (ISBN 91-973090-1-X) Skriftserien ; 11 (nytryck av nr 7, bilaga inne i boken) Barnbibliotek och informationsteknik. Elektroniska medier för barn och ungdomar på folkbibliotek. Red. Anette Eliasson, Staffan Lööf, Kerstin Rydsjö. (ISBN 91-973090-2-8) Skriftserien ; 12 Folkbildning och bibliotek? På spaning efter spår av folkbildning och livslångt lärande i biblioteksvärlden. Red. Maj Klasson. (ISBN 91-973090-3-6) Skriftserien ; 13 Zetterlund, Angela: Utvärdering och folkbibliotek : En studie av utvärderingens teori och praktik med exempel från folkbibliotekens förändrings- och utvecklingsprojekt (ISBN 91-973090-4-4) Slut Myrstener, Mats: På väg mot ett stadsbibliotek. Folkbiblioteksväsendets framväxt i Stockholm tom 1927. (ISBN 91-973090-5-2) Skriftserien ; 15 Limberg, Louise: Att söka information för att lära. En studie av samspel mellan informationssökning och lärande (ISBN 91-973090-6-0, Slut) (ISBN 91-89416-04-X, nytryck 2001 och 2003) Skriftserien ; 16 Hansson, Joacim: Om folkbibliotekens ideologiska identitet. En diskursstudie (ISBN 91-973090-7-9) Slut Gram, Magdalena: Konstbiblioteket : en krönika och en fallstudie (ISBN 91-973090-8-7) Skriftserien ; 18 Hansson, Joacim: Klassifikation, bibliotek och samhälle. En kritisk hermeneutisk studie av ”Klassifikationssystem för svenska bibliotek” (ISBN 91-973090-9-5) Skriftserien ; 19 Seldén, Lars: Kapital och karriär. Informationssökning i forskningens vardagspraktik. (ISBN 91-89416-00-7 Slut) (ISBN 91-89416-08-2, nytryck 2004) Skriftserien ; 20 Edstrom, Göte: Filter, raster, mönster. Litteraturguide i teori- och metodlitteratur för biblioteks- och informationsvetenskap och angränsande ämnen inom humaniora och samhällsvetenskap. (ISBN 91-89416-01-5) Slut Röster. Biblioteksbranden i Linköping. Red. Maj Klasson (ISBN 91-89416-02-3) Skriftserien ; 22 Stenberg, Catharina: Litteraturpolitik och bibliotek. En kulturpolitisk analys av bibliotekens litteraturförvärv speglad i Litteraturutredningen L 68 och Folkbiblioteksutredningen FB 80. (ISBN 91-89416-03-1) Skriftserien ; 23 Edström, Göte: Filter, raster, mönster. Litteraturguide i teori- och metodlitteratur för biblioteks- och informationsvetenskap och angränsande ämnen inom humaniora och samhällsvetenskap. Andra aktualiserade och utökade upplagan. (ISBN 91-89416-05-8) Skriftserien ; 24 Sundin, Olof: Informationsstrategier och yrkesidentiteter - en studie av sjuksköterskors relation till fackinformation vid arbetsplatsen. (ISBN 91-89416-06-6) Skriftserien ; 25 Hessler, Gunnel: Identitet och förändring - en studie av ett universitetsbibliotek och dess självproduktion. (ISBN 91-89416-07-4) Skriftserien ; 26 Zetterlund, Angela: Att utvärdera i praktiken - en retrospektiv fallstudie av tre program för lokal folkbiblioteksutveckling. (ISBN 91-89416-09-0) Skriftserien ; 27 Ahlgren, Per: The Effects on Indexing Strategy-Query Term Combination on Retrieval Effectiveness in a Swedish Full Text Database. (ISBN 91-89416-10-4) Skriftserien ; 28 Thórsteinsdóttir, Gudrun: The Information Seeking Behaviour of Distance Students. A Study of Twenty Swedish Library and Information Science Students (ISBN 91-89416-11-2) Skriftserien ; 29 BO JARNEVING THE COMBINED APPLICATION OF BIBLIOGRAPHIC COUPLING AND THE COMPLETE LINK CLUSTER METHOD IN BIBLIOMETRIC SCIENCE MAPPING Bibliometrics is the quantitative study of patterns derived from the production and use of publications where mathematical and statistical methods are applied. The focus of this thesis connects to previous research in bibliometric science mapping and citation indexing. A citation index facilitates the retrieval of documents associated through citation links whereas the objective of citation based science mapping is to reveal the cognitive structures of science and to provide scientists with information. In the empirical study presented, a method for the partition of document populations, the complete link cluster method and a method for the measuring of similarity between research articles, bibliographic coupling, were combined to a method preliminarily fit for science mapping purposes. The aim of the study was to evaluate this method and to find its area of application. Findings showed that the proposed method has the capability to identify and map current and coherent research themes on the level of a single research field as well as in a multidisciplinary environment. However, based on theoretical considerations as well as on empirical findings, it was concluded that it would not suffice as a standard science mapping method where exhaustive depictions of specialties’ cognitive structures are aimed at. On these grounds, it was concluded that the area of application of the proposed method should be scientific information provision and that it would be complementary to traditional citation indexing. Bo Jarneving is a member of the teaching staff at the Department of Library and Information Science/School of Library and Information Science, University College of Borås and Göteborg University. The Combined Application of Bibliographic Coupling and the Complete Link Cluster Method in Bibliometric Science Mapping is his Doctoral Thesis. bl g Department of Library and Information Science/ Swedish School of Library and Information Science University College of Borås and Göteborg University ISBN 91-89416-12-0 ISSN 1103-6990