摘要
Web中的客户评论信息挖掘是大数据分析中的一项重要内容。分析客户评论中所包含的产品特征情感倾向,不仅可为消费者购买产品提供更具体的决策支持,还能有效帮助企业改进产品质量。针对商业应用的实际需要,提出了一种自动从中文客户评论中抽取产品特征并判断其情感倾向的方案。基于frequent pattern-tree(FP-tree)方法提取产品特征,结合基于语料库的方法和依存句法分析方法识别关于产品特征的主观评论语句、情感词及其情感词的依存关系,综合考虑情感词、否定词、程度词计算产品特征的情感倾向值。采用公开数据中的600篇手机评论作为实验数据,检验了算法的准确性。对比分析的结果说明,算法有很好的应用潜力,能够有效地从网络评论中获取有价值的商业信息。
Data mining in customer reviews of Web is one of the important applications in big data analysis. Extracting semantic orientation on product feature from customer reviews can not only provide customers with more detailed information for purchasing decision,but also help enterprises improve product quality. Aim at the actual requirement of business application,an algorithm of autocratically extracting customers’ semantic orientation on product features from network reviews is proposed. In the proposed algorithm,the product features are obtained based on frequent pattern-tree（ FP-tree） method.Then,the subjective review sentences on product features,emotional words and its dependency rules are recognized according to corpus-based approach and dependency parsing method. Finally,the semantic orientation values of product features are calculated by considering the effect of opinion words,negative words,and degree words. The public data set with 600 mobile phone reviews is used to test the performance and validity of the proposed algorithm. Moreover,the comparing results show that the presented algorithm has the high potential to be applied to extract valuable commerce information from network product reviews.
作者
王永
陶娅芝
张勤
WANG Yong TAO Yazhi ZHANG Qin(Key Laboratory of Electronic Commerce and Logistics of Chongqing, Chongqing University of Posts and Telecommunications Chongqing 400065, P. R. China Computer Engineering Department, San Jose State University, California, USA, 95192)
出处
《重庆邮电大学学报（自然科学版）》
CSCD
北大核心
2017年第1期75-83,共9页
Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)
基金
国家自然科学基金(61472464)
重庆市前沿与应用基础研究项目(cstc2015jcyj A40025)
重庆市社会科学规划管理项目(2015SKZ09)
重庆市社科基金(K2015-59)
重庆邮电大学社科基金(K2015-10)~~
关键词
情感倾向分析
产品特征
语义相似度
WEB挖掘
知识发现
sentiment orientation analysis
product feature
semantic similarity
Web mining
knowledge discovery
