基于主体句和句法依赖的微博情感倾向性分析

情感倾向性分析是近年来中文信息处理领域的热点问题.通过对新浪微博进行情感的分析与研究,提出了一种基于主体句和句法依赖关系的微博情感倾向性分析方法.首先利用自定义规则和条件随机场模型进行主体句及主体评价对象的抽取;然后使用句法分析器对主体句进行依赖关系分析,可以准确的获得修饰评价对象的评价词;最后利用情感词典计算出句子的情感倾向.实验结果表明在精确获取评价对象的基础上再进行情感倾向性判别效果要优于对微博直接进行情感倾向性分析.

Sentiment Analysis of Micro Blog Based on the MainSentence and Syntactic Dependencies

Sentiment analysis is a hot issue in the field of Chinese information processing in recent years.Through the analysis and study of emotion on Sina micro-blog,an analysis method based on the main sentence and syntactic dependencies was proposed.First,system use the custom rules to extract the main sentence,and use the CRFs to extract the opinion targets from the main sentence;Then apply the parser to obtain dependencies from main sentence,which can accurately get the evaluation object words that modify opinion targets;Finally,calculate the sentence emotional tendencies by emotional dictionary.Experimental results show that sentiment analysis on the basis of obtaining opinion targets precisly is better than direct analysis.