基于语义分析的中文微博情感分类方法  

Sentiment classification method of Chinese Micro-blog based on semantic analysis

作　　者：杨佳能[1,2] 阳爱民[3] 周咏梅[3] 

机构地区：[1]广东外语外贸大学国际工商管理学院,广东广州510006 [2]广东省电信规划设计院有限公司,广东广州510630 [3]广东外语外贸大学思科信息学院,广东广州510006

出　　处：《山东大学学报（理学版）》2014年第11期14-21,30,共9页Journal of Shandong University(Natural Science)

基　　金：国家社科基金资助项目(12BYY045);教育部新世纪优秀人才支持计划资助项目(NCET-12-0939);广东省教育厅科技创新项目(2013KJCX0067)

摘　　要：通过分析微博的结构特点,提出了一种基于语义分析的中文微博情感分类方法。首先构建了表情符号情感词典和网络用语情感词典;然后结合词典资源对微博文本进行依存句法分析并且构建情感表达式树;最后根据制定的规则计算微博文本的情感强度,依据强度值判断微博的情感倾向类别。实验结果验证了该方法的有效性,也表明所构建的表情符号情感词典和网络用语情感词典能够有效增强情感分类器的性能。By analyzing the structural features of Chinese Micro-blog,a sentiment classification method based on seman-tic analysis was proposed.For the proposed method,firstly,an emoticons sentiment lexicon and a network language sentiment lexicon were built.Then by using these two lexicons and the dependency parsing results,the sentiment ex-pression Binary Tree was constructed.Finally,the sentiment strength,which was calculated by the established rules, was applied into the sentiment classification.Experimental results show that this method is effective and two built senti-ment lexicons can better enhance the performance of the sentiment analysis system.

关 键 词：中文微博 情感分析 依存句法分析 表情符号 网络用语 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]