Sentiment Tendency Analysis Based on Semantic for MicroBlog

Abstract

With the rapid development of Internet, microblog has become an important tool for online interaction, and sentiment tendency analysis for microblog text also has become a hot spot in the field of information mining. Because of the extremely complex Chinese semantic environment, the ordinary sentiment analysis method based on machine learning is not very good at Chinese microblog. Therefore, this paper proposes a sentiment tendency analysis method based on semantic for Chinese microblog text. We take emotional word set of HowNet as the basis, and then construct a sentiment dictionary for Chinese microblog. After the microblog text is performed pretreatment and feature selection, we calculate the weight of emotional words, get the sentiment tendency by computing the weight sum of microblog text. Experimental results show that this method can more effectively judge the microblog text emotion bias compared with HowNet-Based classification and SVM-based classification.