Research And Application On Chinese Micro-Blog Sentiment Classification

With the rapid development of Internet, Micro-blog as the new information publishing and sharing platform attracted a large number of users to express their opinions, emotions and ideas since the Web 2.0 era. The researches about micro-blog sentiment classification contribute to grasp the views and attitudes of users, control the public opinion, explore the market demand, etc. At present, compared with Chinese microblog, the researches on sentiment classification of English microblog have been mature. However, due to a late start and the Chinese syntax and semantic complexity, the researches on sentiment classification of Chinese microblog are still in the exploratory stage. In addition, there is rarely relevant research realized the microblog sentiment classification system based on a particular filed.Aimed at above issues, this paper based on the research of Chinese microblog sentiment classification, research works have been done based on existing text classification technology as follows.Firstly, this paper researched the data acquisition method of Sina microblog. Analyzed the limitations of API, based on simulating login and parse page, this paper designed and implemented a Sina microblog Web page grab method to get the relevant topic of microblog search page data. It supplied the better data-based for future researches.Besides, based on the applicability research about the traditional sentiment dictionary, this paper integrated two basic emotional dictionary resources. In addition, the microblog emoticons dictionary and the microblog network dictionary were constructed.In addition, this paper studied general method of sentiment classification feature selection, summed up the Chinese microblog basic feature set and added unitary emoticons and unitary vocabulary information in it. This paper constructed a multi SVM classification model to divide microblog into three categories: positive, negative and neutral. The validity of the proposed method and model is verified by experiments.Finally, based on the study of Chinese microblog sentiment classification, this paper designed and implemented a hotel microblog sentiment classification system. The goal of the system is to classify the microblog user's reviews of different hotel, to understand the user brand recognition of different hotel, and to evaluate the brand effect to different hotel.The hotel microblog sentiment classification system confirmed the applicability of the analytical model and method proposed in this paper. The results are conducive to understand the real service level, to further improve the hotel service and reasonable consumer choice for the hotel industry and consumers.