A Network Portrait Divergence Approach to Measure Science-Technology Linkages

Science and technology (S&T) association has become an important pattern of promoting technological innovation. Detecting S&T linkages is a crucial way to discover their knowledge connections. Prior research mostly identifies S&T linkages by counting statistical distribution or calculating the semantic similarity of terms or topics between S&T, while is unequipped to reveal structural linkages between S&T systems. In this study, we proposed a novel knowledge network coupling approach to gauge network linkages between S&T based on a network portrait divergence algorithm. First, we extracted core knowledge elements in S&T and applied a word alignment approach to uniformly descript those elements. Then, by transforming the S&T knowledge system into knowledge networks, we constructed S&T knowledge networks based on word co-occurrence relations. Finally, a network portrait divergence algorithm is applied to generate network portraits and calculate portrait distance between S&T knowledge networks to gauge their dynamic structural coupling. The results demonstrated that the knowledge network-based coupling approach could detect the linkages and interaction patterns of S&T at a fine-grained level. This study can provide methodological tools and reference basis for the development and policymaking of S&T innovation.