Detecting Overlapping Communities in Social Networks by Game Theory and Structural Equivalence Concept

Most complex networks demonstrate a significant property ‘community structure’, meaning that the network nodes are often joined together in tightly knit groups or communities, while there are only looser connections between them. Detecting these groups is of great importance and has immediate applications, especially in the popular online social networks like Facebook and Twitter. Many of these networks are divided into overlapping communities, i.e. communities with nodes belonging to more than one community simultaneously. Unfortunately most of the works cannot detect such communities. In this paper, we consider the formation of communities in social networks as an iterative game in a multiagent environment, in which, each node is regarded as an agent trying to be in the communities with members structurally equivalent to her. Remarkable results on the real world and benchmark graphs show efficiency of our approach in detecting overlapping communities compared to the other similar methods.