Efficient stepwise detection of communities in temporal networks

In temporal networks, dynamic community detection is composed of two separate stages: (i) community detection at each time step; (ii) community matching across time steps. In the traditional methods, the community matching across time steps is based on nodes, which is time consuming. In this paper, we suggest a simple method which takes advantage of historic community information to detect dynamic communities. After dividing each community at previous time step into a few modules, we cannot only use these modules to detect communities at current time step but also map communities across time steps. Results on synthetic and real networks demonstrate that our method cannot only maintain the quality of communities but also improve the efficiency of community matching significantly.