Unlocking trends in secondary battery technologies: A model based on bidirectional encoder representations from transformers

Abstract

Battery technology is widely used in various aspects of modern life, and efficient energy storage is becoming increasingly crucial. Secondary battery technology is continuously developing, and its market value is increasing. Therefore, data analysis is essential for the continued growth of technology in this field. Patent data is commonly analysed to identify technological trends, providing valuable information for technological innovation and competitiveness. Compared to traditional topic modelling techniques based on word occurrence frequency, Bidirectional Encoder Representations from Transformers (BERT) demonstrates superior natural language processing results in generating contextual word and sentence vector representations by considering the semantic similarities of the text. Therefore, this study utilised this model to extract topics. From a total of 6218 patent data, this study extracted core topics and the main keywords for secondary battery technologies between 2013 and 2022 were lithium-ion, electric vehicles, unmanned air vehicles, and solar panels, confirming the accuracy of BERT-based patent analysis. Additionally, this study selected the topics and present their main concepts and trend analysis to provide insights into future research on secondary battery technologies.