Research on the Construction of Sentiment Lexicon Based on Chinese Microblog

摘要

Massive micro-blog's emotion analysis on the Internet is one of the current research hot topics. In view of low coverage of the existing Chinese sentiment lexicon, This paper constructs a basic micro-blog sentiment lexicon by integrating the current sentiment lexicon. At the same time, extract the key problem of Chinese micro-blog sentiment analysis: how to identify the micro-blog new words and explain the emotional meaning, and put forward two methods to judge the sentiment words that not included currently, that is based on the CHI square and improved SO-PMI algorithms which based on associative computing. Finally, the experimental results show that the constructive lexicon can be effectively used for feature selection and classification, and achieves a better classification performance.