Uncovering Green Innovation: A Topic Modeling Analysis of Climate Change Mitigation Patents

Abstract

This thesis examines the thematic structures and innovation maturity of technologies for mitigating climate change by analyzing the descriptive text of these technologies, specifically patent titles and abstracts. This study proposes a novel methodological approach for mapping and interpreting the development of green technology by combining Latent Dirichlet Allocation (LDA) topic modeling with logistic S-curve analysis. The research adopts an exploratory approach, focusing on the CPC classification method called the Y02 tagging scheme, which categorizes technologies for mitigation or adaptation against climate change. A time analysis of patent applications, modeled with S-curves, demonstrated that general energy storage technologies (Y02E60) entered a maturity phase around 2014, while energy storage using batteries (Y02E60/10) continues to experience accelerated growth. Y02E60 was the second most prevalent CPC group, after Y02E10, which categorizes renewable energy sources. Y02E10 exhibited a persistently stagnant trend in patent applications and was consequently rejected for further analysis. This study examines 7,211 patents classified under CPC subclass Y02E60/10. LDA disclosed six coherent subjects, including battery cooling systems, chemical composition, safety modulation, and design, underscoring the multifarious technical challenges inherent in battery innovation. This research not only promotes the application of machine learning and innovation theory in patent analysis but also offers practical guidance for R&D strategists and decision-makers in navigating the changing landscape of green technology. In summary, this thesis demonstrates the value of combining text mining and innovation theory to extract useful insights from patent data, contributing to more informed decisions in the rapidly emerging field of climate change mitigation technologies.