Particle Swarm Optimization For Complex Network Community Detection

Complex network analysis is one of the theoretical foundation of social computing and big data processing, and the analysis and study of complex networks is of great importance. Complex network analysis involves many problems, such as recommendation on networks, the robustness of networks, network resource allocation, network structure information mining, etc., and the network structure information mining should be the basic theory of network analysis, many network problems inevitably need to analyze and mine the structure information so as to guide solving of network problems. This dissertation mainly study the community structure mining of complex networks, from the perspective of swarm intelligence optimization, this paper proposes a feasible complex network community mining method based on particle swarm optimization algorithm. The main work of this paper is as follows:1) Surveys the related research articles on the analysis of complex network community structure, get to know the definition of a community of a network, have a macroscopic overview of the existing approaches for network community detection, get to know the merits and demerits of the existing avenues, and experiments on some algorithms are carried out.2) Surveys the researches on swarm intelligence optimization, with emphasis on particle swarm optimization algorithm. Have a deep understanding of the basic principle of particle swarm optimization algorithm and the current landmark research progress on the particle swarm optimization algorithm.3) In view of the complex network community mining problems, from the viewpoint of optimization, on the basis of deep understanding of the particle swarm optimization algorithm, this paper proposes a feasible discrete particle swarm optimization algorithm for solving complex network community mining problems. In the designing of the proposed algorithm, in order to combine particle swarm optimization and network community detection, this paper redesigns the discrete representation of the particles. In order to enhance population diversity of the algorithm, based on the prior knowledge of network topology, a new particle status update rule is designed. In order to improve the global search ability, a local search strategy is proposed on the basis of the community property.4) In order to verify the effectiveness of the algorithm proposed in this paper, the experiment part adopts a lot of artificial simulated data as well as a large number of real complex networks to test the proposed algorithm. In addition, a existing algorithm proposed in the literature is compared with the proposed algorithm. The experiments demonstrate that the proposed particle swarm optimization based community detection algorithm has good community mining performance, and the designed local search strategy can enhance the global searching ability.