Discovering overlapping communities in ego-nets using friend intimacy

Methods for detecting overlapping communities are essential for understanding complex network structures and extracting implied information. Traditional community detection algorithms have been proven to be unsatisfactory when the network community structure is relatively fuzzy. In this paper, we proposed a novel overlapping community discovery algorithm (ENFI) to address this problem on the micro level using ego-nets. The ENFI approach exploits the micro-characteristics of ego-nets, extracts the ego-net's local community by calculating the friend intimacy, and then forms the overlapping communities of the network. We conducted experiments on both synthetic and real-world social networks using normalized mutual information (NMI) and overlapping community modularity as evaluation criteria. The results demonstrated that the proposed ENFI algorithm can detect community structures in complex networks more efficiently and accurately than existing state-of-the-art algorithms.