With the rapid development of Internet technology, the virtual large-scale social networks are wide- spread. Community structure is an important characteristic of social network, mining the communi-ty structure in large-scale social networks can help us to understand the internal structure and relationships of the network, so as to better apply these networks. Therefore, community detection has important practical significance. A kind of individual rationality as the core of cooperative game community detection algorithm was proposed in this paper, which based on cooperative game com- munity detection model and efficient iterative formula for computing the Shapley value (SH). CDCG algorithm includes initial detection and community adjustment. In the initial detection of changing the strategy environment, every node based on its own maximum SH value to make a decision, after several rounds of decisions, when all of the nodes’ SH value become balance in the network, then the initial detection ended. The characteristics of the internal tight connection of community and rela-tively sparse in communities, by which can detect the unreasonable and meaningless small clusters in step of community adjusting, then the detected cluster has obvious community features. In order to improve efficiency of the algorithm, no contribution nodes pruning and ownership nodes pruning strategies were proposed. Finally, extensive experiments show that CDCG algorithm can automati-cally determine the final number of divided communities, which is effective and efficiency.
Zhao, Z.Y., Feng, S.Z., Wang, Q., Huang, Z.X., Williams, G.J. and Fan, J.P. (2012) Topic oriented community detection through social objects and link analysis in social networks. Knowledge-Based Systems, 26, 164-173.