A Memetic Algorithm for Community Detection in Bipartite Networks

Bipartite networks belong to the category of complex networks, whose vertices can be divided into two separated vertex sets, so that there are no edges between vertices in the same set, and edges only exist between vertices in different sets. In the past decades, although community discovery in one-mode networks has been deeply explored, the detection of communities in bipartite networks has not been widely studied. In this paper, we present a new memetic algorithm named MATMCD-BN for community detection in bipartite networks with two types of nodes in the community. Firstly, we put forward a new initialization method for population initialization of memetic algorithm for bipartite network communities discovery, which can expedite the convergence speed of this algorithm. Secondly, besides using traditional mutation operator, we propose a new crossover operator (called two-way random crossover operator in this paper) and a new mutation operator (called mutation operator 2 in this paper), which are helpful to improve the accuracy of the solution and accelerate the convergence speed of the proposed algorithm. Finally, we develop a local search method, which can make the solution approach the global optimal solution quickly and jump out of the local optimal solution with a certain probability. As far as we know, the proposed MATMCD-BN is the first memetic algorithm (MA) applied to community detection in bipartite networks with two types of nodes in the community. In order to confirm the performance of this algorithm, we have done a lot of experiments using synthetic and real social networks. The experimental results demonstrate that the presented method is effective and promising for bipartite network community identification.