Hierarchical Cooperative Social Spider Algorithm for Community Detection in Large Network

Community structure is one of the important features in networks to understand the complex systems. Detecting community in networks is often formulated as an optimization problem in recent years. The nature-inspired optimization methods have been used to deal with this problem. However, these algorithms are facing challenges when they are applied to solve community detection problems in a large-scale social network. In this paper, we propose Hierarchical Cooperative Social Spider Algorithm (HCSSA), a novel method to reveal community structures in complex networks. The spider statuses are redefined with a discretization scheme. The spider updating mechanism is redesigned based on the network topology. In order to enhance the diversity and quickly search capability of the algorithm, a hierarchical cooperative multi-swarm is applied to divide the spider into several swarms. Especially, a greedy search tragedy is used in the main swarm to speed up the convergence of algorithm whereas the other swarms keep the diversity of the population. Lastly, the proposed algorithm is evaluated on some real-world networks. The experiment results demonstrate that HCSSA has an effective and guaranteed performance for dealing with community detection in complex networks.