Design And Implementation Of An Implicit Microblog Theme Mining System

With the rapid development of the Internet, electronic business platform has become the main channel for public purchase of goods, business official Microblog is a new way to service in electronic business platform facilitates the release of outside news and let users discuss their products. With the increase of network marketing company influence, associated with the company’s growing number of microblog text, which implies some topics that closely related to the companies. Mining get public opinion trends which valuable information can be associated with the company’s real-time monitoring, support the company’s management decisions. Microblog topic detection technology is the study of how a large number of microblogs information management data classification, which has become microblogging study in one of the hottest current direction.The innovations of this paper are as follows:(1) This paper analyzes the structure from the beginning of Microblog. since the text is short, less words, grammatical style random characteristics, therefore contains a large number of microblogs can not analyze its theme attribution. But if you blindly discard it will greatly affect the overall theme of the output, left out a lot of microblogs author topic of concern. Therefore, this article on the microblog topic model currently exists, focuses on the structure of microblog and the relationship between one microblog and other microblog and combine probabilistic topic model to design and propose a new topic model based on implicit microblog by using comment and retransmission microblog and their own context based, which called CGRMB-LDA model.This model use comment group and retransmission relationship,using comment, retraismission and context relationnnship in Microblog to expand implicit microblog,and using Gibbs Sampling implementation for inference of model. Finally model get the result that microblog-topics and themes-vocabulary probability distributions.(2) On the implict topic microblog mining system implementations, the decline was due to consider the use of obvious microblog to expand implicit microblog, therefore need the help of comment and implementation microblog relationship. This article also discusses how to get quickly and easily using microblog open platform API interface to obtain comment and implementation microblog relationship to analyze and deal with implict microblog. In addition, we also consider the data preprocessing for implict microblog, network symbols replace and emotions word expand so that the output is more accurate and with some emotional color.