Metaheuristic Multi-Objective Method to Detect Communities on Dynamic Social Networks

Abstract

Community detection is an important area in social networks analysis, which has many applications. Most social networks are inherently dynamic, consisting of constantly changing communities and therefore, community detection is a challenge in such networks. Since the communities in dynamic networks change, we need high-performance methods for community detection which observe the network changes and update the detected communities, instead of finding the communities for each network snapshot from scratch. As a result, the need to incremental community detection algorithms has been emerged. In this paper, a novel method is presented to identify communities in dynamic social networks, based on a multi-objective metaheuristic algorithm using label propagation technique, in order to detect communities incrementally. The evaluation of the proposed method, which includes examination of different artificial and real networks, shows that the proposed algorithm outperforms the state-of-the-art algorithms with respect to modularity and normalized mutual information (NMI) objectives.

Keywords

• Dynamic networks
• Community detection
• Heuristic algorithms
• Grey Wolf optimize