Overlapping Community Detection Algorithm Based on Adjustable Fitness and Share Neighbors.

Abstract: The current overlapping community detection algorithm doesn’t take the neighborhood communities’ contribution to the neighbor overlapping into account. In this paper, we put forward an algorithm of detecting overlapping communities OCA which base on adjustable fitness and share neighbors. Its basic idea is that local information will help discover natural communities. Firstly, we take nodes with maximum degree as community seed and add neighbor nodes with maximum fitness to it. Then we take nodes that still do not be included in any communities, as seeds of new communities and iterate the community detecting process. Because some detected communities are overlapping, we introduce a quality function COD to evaluate community topology. And we take COD as the standard of whether the overlapping parts should be combined or not. Experiments on synthetic data sets and realworld networks show our method can accurately detect overlapping communities. [Copyright &y& Elsevier]