A novel approach for overlapping community detection in social networks based on the attraction

Abstract

The growing scale of networks makes the study of social networks increasingly difficult. Overlapping community detection can both make the network easier to analyze and manage by detecting communities and better represent the intersection between communities. In this paper, a novel approach for overlapping community detection in social networks is proposed. First, the nodes with local maximum degree are selected from the global network to form initial communities. Next, if the attraction between a community and its surrounding node exceeds a set threshold, these nodes can be directly attracted to that community. Then repeat the above process iteratively until communities no longer change, and nodes that have not yet been divided into communities are regarded as overlapping nodes if they are attracted to two or more communities all greater than the set threshold. In addition, the membership of an overlapping nodes in a related community can be calculated by computing the ratio of the attraction of that community to the overlapping node to the sum of the attractions that the node has. Finally, experimental results on 4 synthetic networks and 6 real-world networks show that the proposed algorithm is effective in detecting overlapping communities and performs better compared to some existing algorithms.