Community Detection Utilizing a Novel Multi-swarm Fruit Fly Optimization Algorithm with Hill-Climbing Strategy

The community detection methods based on evolutionary algorithm have become a hot research topic in recent years. However, most contemporary evolution-based community detection algorithms need many parameters in the initialization process and are characterized by complicated computational processes, which are puzzled for users to have a better understanding of these parameters on the performance of corresponding algorithm. In this paper, we first propose a new community detection method utilizing multi-swarm fruit fly optimization algorithm (CDMFOA), which needs only a few parameters and has a simple computational process. Moreover, we adopt the multi-swarm fruit fly strategy and hill-climbing method in community detection algorithm in order to resolve the premature convergence and improve the local search ability of CDMFOA. Meanwhile, we separately utilize modularity and modularity density as objective function in the framework of the CDMFOA, named CDMFOA_Q and CDMFOA_D, so as to check their detection abilities and accuracies in partitioning communities of complex networks. The experimental results on synthetic and real-world networks show that CDMFOA can effectively detect community structure in complex networks. Besides, we also demonstrate that the CDMFOA_D performs better than CDMFOA_Q and other traditional modularity-based methods.