Dynamic Overlapping Community Discovery Based on Core Nodes

摘要

Social networks in the real world are evolutionary and large scale. Detecting the community structure could express the structure and characteristics of complex networks effectively. Many classic incremental clustering and evolutionary clustering algorithms have been proposed to detect the communities in dynamic networks. However, these algorithms rare to consider the importance of nodes, the overlap between different communities during the process of detection. In this paper, an algorithm based on core nodes was proposed which could not only detect dynamic overlapping communities, but also trace the evolution of network communities. Meanwhile, a three-way representation of a community by a pair of sets is introduced to describe the overlapping communities. Experiment results on real-world data sets demonstrate that our proposed method performs better than the well-known dynamic community detection algorithm.