The role of geographic proximity on citation preferences: The case of Artificial Intelligence

This paper presents a methodology to study geography-based assortative mixing patterns of citations preference, under the assumption that geographic proximity plays a role in knowledge diffusion. We modelled citation relationship as a complex network where scientific publications are represented as nodes and each directed link from a publication to another one models the event that authors of a scientific publication have cited another publication. We then studied assortative mixing pattern defined as nodes’ preference to be connected to nodes that have similar characteristics. We explore individual variation in assortative mixing patterns using data from a publication set on Artificial Intelligence. In this study, geography proximity has been codified as a country-based classification of publications. We focused the scope of the analysis to those citation patterns interplaying between publications by authors affiliated uniquely to a single country. Results suggest that geographic proximity plays a significant role in determining knowledge diffusion preferences, but also individual variation exists within each geography-based group and that preference is either directed toward the same country or systematically toward other few countries. The latter empirical evidence helps to unveil what are the country-level research bases where scientific credit is mostly directed.