A Novel Graph-Based Method to Study Community Evolutions in Social Interactions

摘要

Understanding community evolutions in social interactions is meaningful to study user behavior in social networks. In fact, social interactions have highly dynamic characteristics, and the time when the interactions have occurred is necessary for studying evolutions of communities. However, the temporal information is ignored in existing studies. Therefore, a graph-based method incorporating a decay function is proposed to model social interactions in this paper. Furthermore, the CPMw (Clique Percolation Method with Weights) method is applied to the graph to detect communities in each time slice. In this way, the evolutions of communities could be well investigated. Finally, the Digg dataset is used to evaluate the proposed method. The experimental results demonstrate that the temporal information is useful to study the community evolutions in social interactions.