With the rapid development and a large popularity of microblogging, community discovery has become the current new research focuses, which could help the operators understand network structure and the characteristics of users, then provide users with personalized services. Most of previous study only emphasized the network structure without considering the content. The paper provides a community discovery method based on the users' theme similarity and network structure. Firstly, retrieve users' theme from their microblogging; then cluster the similar users based on the links among them and users' similarity; finally gain the communities. The experiments on the big data show that the method can not only find potential community, but also gain its theme.