Evolution of Scientific Collaboration Network Driven by Homophily and Heterophily (1510.07763v1)

Abstract: Many scientific collaboration networks exhibit clear community and small world structures. However, the studies on the underlying mechanisms for the formation and evolution of community and small world structures are still insufficient. The mechanisms of homophily and heterophily based on scholars' traits are two important factors for the formation of community and inter-communal links, which may deserve further exploration. In this paper, a multi-agent model, which is based on combinatorial effects of homophily and heterophily, is developed to investigate the evolution of scientific collaboration networks. The simulation results indicate that agents with similar traits aggregate to form community by homophily, while heterophily plays a major role in the formation of inter-communal links. The pattern of network evolution revealed in simulations is essentially consistent with what is observed in empirical analyses, as in both cases the giant component evolves from a small cluster to a structure of chained-communities, and then to a small world network with community structure. This work may provides an alternative view on the underlying mechanisms for the formation of community and small world structures, complementary to the mainstream view that the small-world is generated from the combination of the structural embeddedness and structural holes mechanisms.