Extracting technical problems from patent texts to support innovation: Integrating human intelligence with machine learning for electric vehicle battery technologies

Technological advancements are accelerating at an unprecedented rate, making it imperative for companies to maintain their own competitiveness by rapidly developing an understanding of evolving innovation trends. Patents, as rich sources of technological knowledge, contain not only solutions but also valuable information on the technical problems that drive innovation. However, extracting and structuring such problem-oriented information is challenging due to the vast and complex characteristics of patent texts. This study highlights the significance of systematically identifying and analyzing technical problems in patent texts to enhance innovation intelligence. By integrating expert knowledge with machine learning, we propose a methodology to extract and categorize technical problems in such texts, thus enabling the construction of a structured database. The proposed approach was applied to the field of electric vehicle battery technologies. We demonstrate its effectiveness identifying cross-sectoral solutions and discovering potential collaboration partners. The findings underscore the importance of systematically structuring technical problem data to enhance technology intelligence, support R&D decision-making, and facilitate strategic planning. Thus, the proposed methodology can be a valuable tool for supporting innovation.