Research On Weibo Sentiment Classification Of Chinese

With the development of the Internet and the Web2.0, Weibo has become one of the most favorite social networking tool for the internet users. Due to its brief message, convenient release and real-time interaction and many other features, Weibo gains more and more popularity. Users prefer to share information, exchange ideas and emotions through Weibo. Because of the growing influence of Weibo, a large number of scholars are now doing various studies about Weibo. Weibo sentiment analysis is one field of them. Through Weibo sentiment analyzing,Weibo-marketing, branding and public opinion monitoring can be successfully achieved. Many analysis research have done for English Weibo sentiment, while Chinese research is still in its infancy. Because of the great increase of Chinese Weibo users and its influence,the sentiment analysis for Chinese Weibo is particularly important and urgent.Sentiment analysis mainly focus on data mining of emotional tendencies in text. Emotional text can be divided into positive, negative, neutral. Text with positive or negative emotion are also called subjective text.Based on its own characteristics of Chinese Weibo and the existing traditional text-based sentiment analysis method, this paper focus on the research of the emotional analysis of Chinese Weibo. Firstly, a highly credible sentiment Lexicon was built,then through the corpus-based learning method and by the use of2-POSW model,the remained sentiment expressions in the sentence could be extracted.Secondly,under the use of appearing expressions in Weibo and in conjunction with high-credible emotion dictionary,we construct Chinese Weibo sentiment corpus,which not only ensures the scale and accuracy,but also reduces the labor burden. Then,based on the sentiment corpus,features of sentiment words and phrases will be extracted and optimized by the use frequency and information entropy.Combined with the sentiment dictionary feature, punctuation feature,we implement a sentiment classification system for Chinese Weibo.In the subject text classification area,Experimental results show that the F Value in this paper increases7%compared with the traditional method,based on the large-scale sentiment lexicon;In terms of sentiment classification,the method in this paper obtains the highest F Value81.9918%.