Social circle detection based on micro-blogging topic and user follow relationship

As the detection of social circles can help the users find the other users with similar interests in a big data environment to expand their friend circles, our algorithm takes the (implicit) user topic in micro-blog and the (explicit) follow relationship between the users into comprehensive account. Firstly, use the supervised-LDA model to extract user topics from micro-blogging data and calculate the similarity between users integrating the follow relationship between them. Then choose the initial clustering center of social circle to cluster user nodes and merge those social circles with a high overlapping degree to detect a social circle finally. The experiment proves that our social circle detection algorithm can not only be used to detect the unknown social circles, but also mine the topics of social circles. Also the comparison with the traditional social circle detection algorithms shows that our algorithm is more effective.