Social Influence and Role Analysis Based on Community Structure in Social Network

摘要

Recent graph-theoretic approaches have demonstrated remarkable success for ranking networked entities, including degree, closeness, betweenness, etc. They are mainly considering the local link factors only, while not so much work concentrates on the social influence ranking based on the local structure in social network. In this paper, two new social influence ranking metrics, InnerPagerank and OutterPager-ank are proposed based on the concept of modified Pagerank, by considering the community structure knowledge. It is well adapted to direct and weighted networks also. Using the two metrics, we also show how to assign community-based node roles to the nodes, which is an effective supplement for single metric used as social influence measure. Identifying and understanding the node's social influence and role is of tremendous interest from both analysis and application points of view. This method is shown to give rasonable results than previous metrics both on test networks and real networks.