The Microblog-based Theme Community Discovery

Microblog, as a new rising web platform for information exchange and sharing, has been frequently refered in a huge number of people’s daily life. Microblog site such as Twitter, Facebook, Sina as well as Tencent have become the containers of massive amount of information. As a matter of fact, some valuable information are stored in those places waiting to be discovery.Although traditional content-based and text-based data mining methods have gained general researches and successes, when apply it to mircoblog, where messages not only precise dialect but also contain social networking relationship, are far from perfect and hardly to be modeled.In order to satisfy market promotion needs to find placement targets, this paper propose an new method to find theme community in microblog, which integrate some potential relation of microblog messages and its social linking network. Authority discovery, text classification, maxflow model and web spider cutting strategies have been creatively applied into an effective and efficient spider.The data in experiment comes from real collection, some analysis have been discussed from several dimensionalities...