Abstract: SNS (social networking service) has been integrated into public life. People upload their own information to the network and use the social networking sites to manage their social relationship. A large number of personal information is presented on social networks. Based on the Twitter platform, this paper designs to ifnd community structure of Twitter users. With the real-time information collecting, this paper rebuilds the relationships of users, improving Fast-Newman algorithms in dividing social networks. The paper uses a visualization system to automatically visualizing community structures, providing decision makers with user network behavior to implement personalized recommendation.