The Implications for Network Structure of Dynamic Feedback Between Influence and Selection

摘要

Although network structures shape diffusion and ultimately systemic performance, the underlying dynamics generating different network structures during diffusion are not well understood. To explore these dynamics we present a set of models in terms of a common drive for relational balance - the inclination to align the attributes of one's network members with one's own attributes either by adopting the attributes of network members as in the diffusion process or by selecting to interact with similar others. Agent based models show that the models generate modular (clustering) structure for high levels of relational balance. Moreover, when network members have strong influence over one another the transition to modularity is delayed but then drastic, in the extreme creating a phase transition. Thus, the rate of influence amplifies the attractiveness of a particular modularity state of the system.