摘要
[目的/意义]针对产品评论中的复合句式,实现特征观点对的语义匹配及提取,并明确评论可信度的识别因素及权重,对产品可信评论进行筛选和分析。[方法/过程]基于特征观点对的语义匹配算法实现评论语义指标的量化计算,并采用模糊层次分析法确定可信度指标权重。[结果/结论]实验表明相较于单句提取特征观点对方法,特征观点对的语义匹配算法在召回率、准确率和F-score等性能方面均有较大优势。依据可信度指标对网站产品评论进行筛选,不仅可以评估产品整体的评论可信度,还可以细化到产品特征级别的可信度分析,为用户筛选可信的评论信息并提升购物决策效率。
[Purpose/Significance]In view of the compound sentence pattern in the product reviews,this paper realized the semantic matching and extraction of the feature opinion pairs,and made clear the indicators and weights of the reviews credibility so as to select and analyze the trusted reviews of the products.[Method/Process]Based on semantic matching algorithm of feature opinion pairs,we extracted the feature opinion pairs and calculated the semantic indicator of reviews,then used Fuzzy Analytic Hierarchy Process to determine the weight of indicators.[Result/Conclusion]The experiment showed that semantic matching algorithm of the feature opinion pairs had a great advantage on the performance of the recall,accuracy and F-score,compared with the method of extracting feature points from the single sentence.It could not only evaluate the credibility of the overall review of the product,but also could be refined to the reliability analysis of the product feature level.Meanwhile,it could screen credible reviews for users and improve the efficiency of shopping decisions.
作者
郝玫
马建峰
Hao Mei;Ma Jianfeng(Donlinks School of Economics and Management,University of Science and Technology Beijing,Beijing 100083,China)
出处
《现代情报》
CSSCI
2019年第6期102-110,141,共10页
Journal of Modern Information
基金
北京市社会科学基金项目"基于客户评论可信度的北京市网络购物评价体系及管理机制研究"(项目编号:17GLC061)
"北京市科技资源错配与创新系统效率提升问题研究"(项目编号:16LJB002)
关键词
产品评论
评论可信度
特征观点对
语义匹配
评论筛选
product reviews
reviews credibility
feature opinion pair
semantic matching
reviews screening
