Community Detection Algorithm For Complex Networks Based on Information Entropy Similarity

Abstract

Community structure is one of the most common and important topological properties of complex networks and community detection provides crucial significance to the study of network characteristics. In the paper, incorporating the information entropy theory, the similarity between nodes and community is redefined, and a heuristic community detection algorithm is proposed, which is called Community Detection Algorithm Based on Information Entropy Similarity (referred as IES algorithm for short). Experimental results demonstrate that the proposed IES algorithm outperforms three traditional methods GN, Fast Newman and CPM algorithm on the karate clubs networks, dolphin social networks and football networks: the modularity value $Q$ of IES algorithm is obviously higher than the other three algorithms, the time complexity and space complexity of IES algorithm are lower than both GN and Fast Newman algorithms, and the efficiency of community partition is also higher. Compared with IE algorithm which is also based on information entropy similarity, the experimental results on Scientists Collaborate Networks further prove that the IES algorithm is more reasonable in the convergence and the community detection rationality.