Group similarity based algorithm for network community structure detection

Abstract

Community structure has an important influence on the structural and dynamic characteristics of the complex system. In the present study, a group similarity model is proposed for the measurement of similarity between two communities. So it can help us understand the mechanism of inter action between these communities. Moreover, based on this model, a hierarchical clustering based algorithm for network community structure detection is put forward. By this algorithm, one pair of communities with the largest similarity is merged in each iteration. And then an evaluation function is adopted for choosing the optimal partition. The algorithm gives a higher performance than many state-of-the-art community detection algorithms when tested on a series of real-world and synthetic networks. Especially, it performs better when the edge density of the network is high.