微博产品评论的情感倾向性分析方法  

Method of Sentiment Orientation Analysis for Microblogging Product Reviews

作　　者：张圣声 阳爱民[1] 周咏梅[1] 杨佳能[2,3] 

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

出　　处：《山西大学学报（自然科学版）》2015年第2期215-222,共8页Journal of Shanxi University(Natural Science Edition)

基　　金：国家社科基金(12BYY045);教育部人文社会科学研究青年项目(10YJCZH247);教育部新世纪优秀人才支持计划(NCET-12-0939);广东省科技计划项目(2010B031000014);广东外语外贸大学研究生科研创新项目

摘　　要：提出一种微博平台上的产品评论情感倾向性分析方法,对特定语料库的产品特征词的词性进行分析,得到特征词的词性重要程度顺序;以情感词典识别的情感词为起点,提出一种依据词性重要程度的"特征观点对"识别规则,并依据该规则进行语义倾向的情感计算。实验结果表明,该方法在产品特征抽取及语义倾向的判断上具有较好的结果。结合微博的转发数和评论数,进行用户观点分析,计算用户对产品的认可度,挖掘产品属性的优缺点,实证分析证明这种方法的可行性。The sentiment orientation analysis method in Microblogging platform is introduced.The order of importance of part of speech is derived from analyzing of part of speech of product character word in specific corpus.And starting with recognition of the sentiment word based on sentiment dictionary,a new identification rule for feature-opinion pairs based on the order of part of speech is proposed,and the semantic of sentence is calculated according to the rules.The experiment shows that this method can get a good result in extracting of character of product and computing of semantic orientation of the corresponding of reviews.Combined with forwarding number and comments number of Microblogging,the analysis of users＇ viewpoint in product reviews will be carried out,which calculate the recognition of products,mine the advantages and disadvantages of product attributes.The empirical analysis proves the feasibility of this method.

关 键 词：微博 产品评论 情感分析 观点挖掘 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]