A density based link clustering algorithm for overlapping community detection in networks

Overlapping is an interesting and common characteristic of community structure in networks. Link clustering method for overlapping community detection has attracted a lot of attention in the area of social networks applications. However, it may make the clustering result with excessive overlap and cluster bridge edge and border edge mistakenly to adjacent communities. To solve this problem, a density based link clustering algorithm is proposed to improve the accuracy of detecting overlapping communities in networks in this study. It creates a number of clusters containing core edges only based on concept named as core density reachable during the expansion. Then an updating strategy for unclassified edges is designed to assign them to the closest cluster. In addition, a similarity measure for computing the similarity between two edges is presented. Experiments on synthetic networks and real networks have been conducted. The experimental results demonstrate that our method performs better than other algorithms on detecting community structure and overlapping nodes, it can get nearly 15% higher than the NMI value of other algorithms on some synthetic networks.