Abstract In order to solve the problem of massive pieces of information on microblogs, this paper studies the centralization theorybased hotspot discovery methods for microblogs, in consideration of the features of microblogging content such as short text, variety of sources and diverse means of dissemination. Through the structured metadata acquired from open APIs, some metadata models for microblogging content are analyzed, and the hotspot discovery process is regarded as a valueadded process of the original materials to clusters of hot products. For initial and deep processing methods during the production process, some data preprocessing techniques as well as short text clusteringbased and disseminating path and users behaviorbased centralizing techniques are proposed. And a complete production and processing model is established. Finally, a series of experiments have verified the theoretical achievement.
2012, Vol. 9 
