Collaborative Community Detection in Complex Networks

Abstract

A collaborative evolutionary model is proposed to address the community structure detection problem in complex networks. The discovery of commmunities or organization of nodes in clusters (with dense intra-connections and comparatively sparse inter-cluster connections) is a hard problem of great importance in sociology, biology and computer science. Based on a natural problem-specific chromosome representation and fitness function, the proposed evolutionary model relies on collaborative selection and best-worst recombination to guide the search process efficiently towards promising solutions. The collaborative operators take into account information about an individual line best ancestor, global and worst individuals produced up to the current generation. The algorithm is able to detect non-overlapping communities in complex networks without the need to a-priori know the expected number of clusters. Computational experiments on several real-world social networks emphasize a good performance of the proposed algorithm compared to state-of-the-art models.