Overlapping Local Community Detection in Directed Weighted Networks

Community detection is an important way to analyze and understand the real networks. In this paper, we not only define Local community modularity and Tightness between local communities for directed weighted networks, but also realize a distributed algorithm that detects overlapping local community in networks. The algorithm is divided into two parts, initial local community detection and similar communities merging. The core of the algorithm is to agglomerate node which causes the greatest local modularity increments for local community, and by iteratively to merge similar communities that have the maximum tightness. Experimental results in real networks prove that the algorithm is reliable.