A spiderweb model for community detection in dynamic networks

Community detection in dynamic networks is one of the most challenging tasks in the field of network analysis. In general, networks often evolve smoothly between successive snapshots. Therefore, the community structure detected in each snapshot should not only be of high quality but also reflect the smoothness of the variations compared with the previous snapshot. In this paper, we propose a novel incremental community-detection method named Spiderweb, which detects the community structure in each snapshot by simulating the evolution of spiderwebs. We categorize the evolutionary events of the network into three types, and then address the changed nodes and edges according to three corresponding evolution rules. In this procedure, some nodes are assigned to proper communities. Then, we construct a new subgraph for the unclassified changed nodes, and detect its communities efficiently. Finally, we merge some communities to obtain the resulting community structure. We conduct extensive experiments on both artificial networks and real-world networks to test the proposed method, and the experimental results show the superiority of the proposed method over some state-of-the-art algorithms in terms of both the quality and the temporal smoothness of the detected community structures. The proposed method provides us with a stable and promising solution for the problem of community detection in dynamic networks.

Automated summary

The proposed Spiderweb method simulates the evolution of spiderwebs to efficiently detect community structures in dynamic networks, achieving superior quality and temporal smoothness compared to state-of-the-art algorithms. It provides a stable and promising solution for the challenging task of community detection in dynamic networks.