Node attraction-facilitated evolution algorithm for community detection in networks

Network model recently has become a popular tool for studying complex systems. Detecting meaningful natural groups of nodes called communities in complex networks is an important task in network modeling and analysis. In this paper, the automatic network community detection is formulated as an optimization problem facilitated by node attraction. The basic idea is envision a network as a system of nodes where each node is attracted by its local neighbors. An evolution community detection algorithm is introduced, which employs a metric, named modularity Q as the fitness function and applies node attraction and modularity-based grouping crossover operator. The proposed algorithm faithfully captures the natural communities with high quality. Node attraction is easy to use for the speed up of the convergence of evolution algorithm to better partitions and for making the algorithm more stable. Node attraction does not require any threshold value. Experiments on synthetic and real-world networks further demonstrate the effectiveness of the proposed approach.