Untangling orientations and dynamics of science–technology interactions through a community-based linkage approach

While many studies have investigated science and technology (S&T) interaction, the fine-grained interaction patterns at the structural level remain unclear. This study proposes a novel community-based linkage approach to elucidate the orientations and dynamics of S&T community interactions. We establish S&T community linkages through network modelling and community detection algorithms, and then quantify the interaction strength, direction and dynamics between different S&T communities. Through an analysis of 790,000 academic publications and 140,000 patents in the artificial intelligence (AI) domain, we find that S&T interaction in this field has continuously strengthened over time. By exploring the structural conditions under which strong S&T community linkages occur, we discover that intensive S&T interactions are more likely to happen within communities of similar size or density. Furthermore, fine-grained differences exist in the science drives technology and technology drives science modalities within AI. This study provides new insights into potential patterns of S&T interaction from a community-linkage structural perspective.