基于基础词典扩展的中文酒店评论情感分析

Sentiment Analysis of Chinese Hotel Reviews Based on the Extended basic Dictionary

针对中文酒店评论自身特点设计语料特征,将评论高频词赋予权重并扩展基础情感词典;结合扩展基础情感词典和语义规则,计算情感加权值,实现对酒店频率褒贬倾向分析;选取Boson和大连理工情感词典作为基础情感词典进行了试验。试验结果表明,利用本方法进行中文酒店评论情感分析的精准率可达到90%以上,相比基础情感词典,可提高10%,且加入前50个高频词扩展基础情感词典,对精准率有较大提升,之后精准率的提升速度趋于平缓。

According to the characteristics of Chinese hotel reviews, the corpus features are designed, the high frequency words of reviews are weighted and the basic emotion dictionary is extended. Then, based on the extended basic emotion dictionary and semantic rules, the weighted value of sentiment is calculated to realize the analysis of the appraisal tendency by selecting the Boson and Dalian Sentiment Dictionaries as the foundation of the sentiment dictionary. The test results show that using the method presented in this paper to conduct sentiment analysis of Chinese hotel reviews, the accuracy rate can reach more than 90%. Compared with the basic sentiment dictionary, the accuracy rate can be increased by 10%, and the top 50 high-frequency words are added to expand the basic dictionary, which greatly improves the accuracy rate, and then the accuracy rate increases gradually.