Sentiment Classification for Topical Chinese Microblog Based on Sentences' Relations

Sentiment analysis is widely applied in product reviews, movie reviews, Twitter and Microblog. In this paper, we throw light on sentiment classification of topical Chinese Microblog, namely, analysis sentiment express style of Microblog, and then classify Microblog to positive, negative or neutral according to sentiment of Microblog. Moreover, the state-of-the-art methods classification sentiment of Microblog by take Microblog as an entirety and ignore sentences' relations (i.e., contrast). Because most of Chinese Microblog have several sentences and these sentences' sentiment is ambiguous even more contradictory, so it is important to consider sentences' relation in sentiment classification. In this paper, we solve the problem by two steps, first of all, we construct the sentiment lexicon, and then we analysis feature for single sentence's sentiment classification. Secondly, we take sentences' relations to optimization sentiment classification result. The experimental results demonstrate our method effectively in Chinese Microblog sentiment classification.