Abstract: Social network is an extension of realistic society in cyberspace. The research on structural characteristics of social network has an important significance on network architecture discovery, network behavior forecast and network security protection. The community structure is one of the basic and important structural characteristics of social network. In recent years, a lot of algorithms for community detecting in social network have been proposed. But they always focuse on unweighted networks, and can’t handle the more and more complex connect relationships between nodes. In order to measure the connection strength in directed and weighted networks, a new definition of node intimacy is proposed. Then, a community detecting method based on node intimacy and degree (CDID) is designed. This method is verified through a series of experiments on synthetic datasets and real-world social network datasets. Compared with other state-of-the-art algorithms, this methed can obtain more accurate community division results under a reasonable run time. And it also provides a unification community detecting method for the four different type networks, such as undirected-unweighted, directed-unweighted, undirected-weighted and directed-weighted networks.

Key words: node intimacy, node degree, weighted networks, modularity, community detecting