A Study On Emotional Tendency Of Chinese Microblogging Based On Conditional Random Field And Emotional Dictionary

With the advent of the 21 st century, the traditional methods of information dissemination has been unable to meet the needs of a new era of information transfer. The rapid rise of the Internet, has promote the traditional information media to new media conversion information from the past behind, monotonous, gradually turning interconnection, active, fast. Internet provides users with a large number of open information exchange tools such as BBS, microblogging, blog, etc. In compliance with the laws of the premise, Users can not restricted by time and space, on a variety of platforms share ideas and spread information, promote the exchange of information through the network unlimitly. Microblog, one of the most representative of the information dissemination platforms, with its fast, convenience, fun,has been accepted and used widely. More and more Internet users started use microblog share informations and express their views, and these views are often with users’ positive or negative emotional color. Weibo has many advantages: large user number, news spread fast and a wide range of topics, therefore, the microblog emotion information mining and emotional orientation analysis has very important significance.In this paper, a number of scholars about text emotional analysis related work is described in detail and summary. Through the Chinese word segmentation, POS tagging, syntactic analysis, this paper introduced the relevant technology of the Chinese text processing in emotional propensity analysis, as well as current common emotional lexicon- Word Net, How Net and so on. Based on the corpus which was provided by Xin Lang Weibo, the article analyzes the problems that facing in Chinese microblog text sentiment analysis.Article also proposed the use of CRFs(Conditional Random Fileds, CRFs) machine learning method and combining emotional dictionary to analysis microblog. Experimental results show the proposed method is avaliable. Experimental results show that compared to the syntactic analysis, the article proposed method improves the accuracy of about 5%. Finally, the article summarizes the lack of analysis of the method and discusses the future emotional analysis techniques. The article also pointed out the directions and the starting points for future work.