Abstract The local optimization algorithm using seed set to find overlapping communities has become more and more a significant method, but it is a great challenge how to choose a good seed set. In this paper, a new method is proposed to achieve the choice of candidate seed sets, and yields a new algorithm to find overlapping communities in complex networks. By testing in real world networks and synthetic networks, this method can successfully detect overlapping communities and outperform other state-of-the-art overlapping community detection methods.