Multi-resolution Community Discovery From Signed Networks Based on Novel Particle Swarm Optimization

There commonly exist friendly and hostile relationships between the individuals in the social networks. The signed network modeling of the social network is one of the effective tool for analyzing the properties of social networks. Recent years, community feature has been proved to be an important property of complex networks. To discover the community structure from signed social networks is of great importance to promote the harmonious development of the society. The task of community discovery from signed networks was modeled as an optimization problem, a novel particle swarm optimization algorithm was proposed to solve the modeled problem. The algorithm optimized a newly suggested objective function called signed link density, which takes a control parameter. By alerting the parameter, the algorithm could obtain the community structures of a network under different resolutions. In order to enhance the global optimization ability of the particle swarm optimization algorithm, a neighborhood dominance based local search operator was designed. To check the performance of the proposed algorithm, experiments on synthetic and real- world signed networks had been carried out, and comparisons with a method existed in the literature had been made. The experiments have demonstrated the effectiveness of the proposed algorithm.