Local Optimization Algorithm Based on Seed Nodes and Belief Propagation for Community Detection

Community structure is one of the intrinsic properties of complex networks and the study of community detection has captured more and more attention from researchers. In this paper, a new community detection algorithm based on seed nodes and belief propagation is proposed. To be specific, we design an improved K-shell algorithm to select seed nodes, each of which is considered as a community root. Then, we expand a community from its root using the belief propagation algorithm. In the post-processing stage, we re-divide the community structures based on the strategy of maximizing modularity and community similarity to further optimizes the quality of community division. Experimental results on different types of networks verify the effectiveness of the proposed algorithm.