Dual-channel hybrid community detection in attributed networks.

This study considers the problem of hybrid community detection in attributed networks based on the information of network topology and attributes with the aim to address the following two shortcomings of existing hybrid community detection methods. First, many of these methods are based on the assumption that network topology and attributes carry consistent information but ignore the intrinsic mismatch correlation between them. Second, network topology is typically treated as the dominant source of information, with attributes employed as the auxiliary source; the dominant effect of attributes is seldom explored or indeed considered. To address these limitations, this paper presents a novel Dual-channel Hybrid Community Detection (DHCD) method that considers the dominant effects of topology and attributes separately. The concept of transition relation between the topology and attribute clusters is introduced to explore the mismatch correlation between the two sources and learn the behavioral and content diversity of nodes. An extended overlapping community detection algorithm is introduced based on the two types of diversity. By utilizing network attributes, DHCD can simultaneously derive the community partitioning membership and corresponding semantic descriptions. The superiority of DHCD over state-of-the-art community detection methods is demonstrated on a set of synthetic and real-world networks. [ABSTRACT FROM AUTHOR]