基于隐朴素贝叶斯的商品评论情感分类方法  

Hidden Nave Bayes based sentiment classification approach of product reviews

作　　者：罗慧钦 陆向艳[1,2] 张雄宝 刘峻[1,2] 

机构地区：[1]广西大学计算机与电子信息学院,广西南宁530004 [2]桂林电子科技大学广西信息科学实验中心,广西桂林541004

出　　处：《计算机工程与设计》2017年第1期203-208,共6页Computer Engineering and Design

基　　金：广西科学研究与技术开发计划基金项目(桂科能1140008-3B);广西高等学校科学研究基金项目(KY2015YB008);广西大学科研基金项目(XJZ130355)

摘　　要：针对朴素贝叶斯模型属性间条件独立假设不完全符合实际的问题,提出一种基于隐朴素贝叶斯模型的商品评论情感分类方法。利用隐朴素贝叶斯分类模型中的属性依赖关系表示商品评论文本中词项间的依赖关系;根据文本词性的语法修饰关系,在计算隐朴素贝叶斯分类模型中的隐藏父节点时去除不具有依赖关系的属性对,仅保留有依赖关系的属性对的计算;将文本属性独立特性加入隐朴素贝叶斯模型中,综合考虑文本属性的依赖性和独立性两个因素对情感分类的影响。实验结果表明,该方法的查准率和查全率优于其它基于朴素贝叶斯模型的情感分类方法。In view of the fact that the conditional independence assumption of Naive Bayes model is not completely consistent with the actual problem, a sentiment classification approach based on hidden Nafve Bayes （HNB） was proposed. The attribute dependency relationship in hidden Naive Bayes classification model was used to express the dependency relationship among the attributes in the product review text. According to the grammatical modification of text part of speech, pairs of attributes without dependency relationship were removed while the values of pairs with dependency relationship were calculated in the calculation of hidden parent nodes of hidden Naive Bayes model. The independent character of text attribute was added into hidden Naive Bayes model, so the impacts on sentiment classification of both the dependency relationship and the independent character of text attributes were considered comprehensively. Experimental results show that the precision and recall rate of the proposed approach are better than that of other sentiment classification approaches based on Naive Bayes model.

关 键 词：隐朴素贝叶斯 商品评论 情感分类 属性依赖 属性独立 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]