Community Detection Based on Weighted Networks

An important property of complex networks is community structure. Community detection is significant to understand the network structure and analyze the network properties. In recent years, lots of algorithms have been developed to find community structure in complex networks. These algorithms, however, are based on unweighted networks, which limits their applications. An unweighted network is only a qualitative description whether there is a connection between vertex pair, so the simple analysis of topological structure cannot depict correctly the structural characteristics of the network. But most real-world networks are weighted ones. Weighted link, whose distribution has a great effect on the property and function of a network, provides a more meticulous depict than unweighted. In this paper, we propose another community detecting algorithm taking into account weights of links. It turns to be especially suitable to the analysis of social and information networks. When tested on both computer-generated and real-world networks, it gives excellent results.