A Community Detection Algorithm Using Random Walk.

Community structure plays an essential role in analyzing networks. Various algorithms exist to find the community structure that scores high on a graph clustering index called Modularity. In divisive community structure algorithms, initially, all the nodes belong to a single community. Each iteration divides the nodes into two groups, and finally, each node belongs to a single community. The main disadvantage of a divisive algorithm is that it is not able to find whether to divide the community further or not. A divisive community detection algorithm is proposed based on the graph spectra that give the termination method for community detection. We rely on Weighted Spectral Distribution (WSD) to divide the network into small sub-network or not. Experiments with various real-world networks show that the proposed method constantly compares favorably with the popular Girvan Newman’s community detection algorithm.