基于商品特征关联度的购物客户评论可信排序方法

发布时间：2019-05-08 03:30

【摘要】：电子商务网站中,海量无序的用户评论可能导致消费者客户"迷失"其中,无法识别评论的可信和真假。针对这个问题,提出了一种根据用户评论的可信度对其重新排序的方法。首先,针对网站商品广告信息,关注在线用户评论内容是否和商品功能属性密切相关,设计了基于HTML脚本格式的购物网站中商品关键特征提取算法,给出了基于自然语言处理的用户评论特征词提取方法;然后,利用词语相似度来分析商品特征和用户评论内容之间的关联度,提出了购物客户评论的可信度计算方法;最后,通过实例分析,实现了大量购物客户评论的可信排序,使得用户无须浏览全部或者大部分之后就能判断哪些评价可以信任或者具有实际的参考价值,降低了信息搜索成本,提高了决策效率。
[Abstract]:In e-commerce websites, the mass of disordered user comments may cause consumers to "lose" them, and can not identify the credibility and authenticity of the comments. In order to solve this problem, this paper proposes a method to reorder user comments according to their credibility. Firstly, focusing on whether the online user comment content is closely related to the product function attribute, this paper designs the key feature extraction algorithm of the shopping website based on HTML script format, aiming at the product advertisement information of the website. The method of extracting user comment feature words based on natural language processing is presented. Then, using the similarity degree of words to analyze the correlation between the characteristics of goods and the content of user's comments, and put forward the method of calculating the credibility of shopping customers' comments. Finally, through the analysis of examples, the trusted ranking of a large number of shopping customers' comments is realized, so that users can judge which evaluation can be trusted or have practical reference value without having to browse all or most of them. It reduces the cost of information search and improves the efficiency of decision-making.
【作者单位】：同济大学计算机科学与技术系;国家高性能计算机工程技术研究中心同济分中心;
【基金】：国家863计划项目(2009AA012201) 国家自然科学基金资助项目(61272107,61202173,61103068) 上海市优秀学科带头人计划项目(10XD1404400) 教育部网络时代的科技论文快速共享专项研究课题项目(20110740001) 华为创新研究计划项目(IRP-2013-12-03)
【分类号】：TP393.092;TP391.1

【相似文献】