Identifying technology opportunity using dual-attention model and technology-market concordance matrix

To understand the role of new technologies in innovation, it is crucial to develop a methodology that links technology and market information. Conventionally, the relationship between technology and the market has been analyzed using a technology-industry concordance matrix, but the granularity of market information is confined by industrial classification systems. In this study, we propose a new methodology for extracting keyword-level market information related to firms' technology. Specifically, we developed a dual-attention model to identify technical keywords from firms' websites. We then vectorized the market information (extracted keywords) and technology information (patents) using word embedding to construct technology-market concordance matrices. Matrices were generated based on a group of high-growth companies to suggest new technologies and market opportunities in the automotive, electronics, and pharmaceutical industries. Finally, two novel indicators are introduced to demonstrate the model's capability in identifying opportunities at the company level.