Abstract: In our report, a novel algorithm for discovering local communities in networks was proposed, which was based on fitness and point similarity, combined the advantages of point community and edge community, and mininged the edge community structure using local information. According to the specific central principle, an original edge was used as a seed. In order to obtain the community structure the local community, the fitness function was constantly maximized, and was used to obtain a local edge community. The modular degree function based on point similarity was used for the boundary node identification.