An incremental community mining method in dynamic social networks

Community mining in social networks can produce meaningful information, such as activity pattern between individuals and the law of social development. Traditional methods for community identification in static social networks may not find the variation of networks. Besides, a few methods on modeling and analyzing community structures in dynamic social networks fail to identify large networks in acceptable time. Therefore, incremental methods to identify community structures in dynamic social networks are proposed to reduce time complexity. However, some of them merely take network topology into consideration, ignoring a large number of attribute information in real social networks. This paper proposes an incremental method to reveal the actual community structure based on attribute weighted networks. In the method, we associate attribute information with the topological graph. Moreover, time complexity is reduced by setting a threshold which represents a reasonable change rate of edge weight. Experiments on a real-world dataset demonstrate that this approach can reduce time complexity and produce nice community structure.