A Cooperative Game Theory-Based Algorithm for Overlapping Community Detection

Community detection is of great significance and has more guiding for network analysis which has attract a lot of attention and is widely used in many areas. The process of community formation is modeled as a game in this paper, nodes in networks are taken as players, and communities are regarded as coalitions. An overlapping community detection algorithm is presented based on cooperative game theory. Furthermore, an edge weight computation is proposed for calculating Shapley value of node and coalition, and a strategy of optimal income for the individual is designed. It is for each player to choose its best coalition to stay or leave during the game process. Through choosing coalitions which can improve the value of their characteristic function, coalitions can be merged to avoid small size coalitions. It does not need any prior information about the number and the size of communities and the overlapping community structure is stable at the end of the algorithm. The experimental results both on synthetic and real networks demonstrate the good performance of the proposed method. It can detect overlapping communities more accurately.