A fast algorithm for community detection in temporal network

Many complex systems can be investigated using the framework of temporal networks, which consist of nodes and edges that vary in time. The community structure in temporal network contributes to the understanding of evolving process of entities in complex system. The traditional method on dynamic community detection for each time step is independent of that for other time steps. It has low efficiency for ignoring historic community information. In this paper, we present a fast algorithm for dynamic community detection in temporal network, which takes advantage of community information at previous time step and improves efficiency while maintaining the quality. Experimental studies on real and synthetic temporal networks show that the CPU running time of our method improves as much as 69% over traditional one.