A graph clustering algorithm based on weighted shared neighbors and links

Community structure is an important topological property of complex networks. Many algorithms have been developed to detect communities in the past, most of them focus on connectivity or attributes of vertices. In the latest ISNGC algorithm, both shared neighbors and connectivity between vertices have been considered. However, it is designed primarily for unweighted networks and has poor performance on weighted networks. To solve this issue, in this study, we propose an efficient clustering algorithm called WSNGC that is based on the weighted shared neighbors and links. The general knowledge is that vertices in the same cluster have more shared neighbors than that in different clusters. We apply our method to some randomly generated networks and compare it with ISNGC algorithm. The experimental results show that our proposed algorithm has good performance on community detection.