基于知识图谱的在线商品评论可信性排序研究  

Credibility Ranking Model for Online Product Reviews Based on Knowledge Graph

作　　者：王忠群 叶安杰 皇苏斌 陈云霞 Wang Zhongqun

机构地区：[1]安徽工程大学管理工程学院,安徽芜湖241000

出　　处：《情报理论与实践》2020年第8期134-139,共6页Information Studies:Theory & Application

基　　金：国家自然科学基金项目“C2C市场中基于行为树的销量识别与发布研究”(项目编号:71371012);教育部人文社会科学研究规划基金项目“基于知识图谱的网购商品评论可信性模型与评估方法研究”(项目编号:18YJA630114)的成果。

摘　　要：[目的/意义]从海量无序的商品评论中排序出可信的评论对消费者网购决策具有重要意义。[方法/过程]首先在给出评论可信度概念后对知识图谱进行赋权扩展,以领域常识、商品测评和领域专家作为信息源运用带权知识建立刻画商品本质信息的带权知识图谱。然后,提取评论中包含商品本质信息的知识以及与商品本质信息相一致的特征观点对来构建评论的语义网。再以评论语义网作为载体计算商品评论的可信度并实施排序。[结果/结论]模型能够更准确地排序出可信的商品评论,方便消费者获取网购决策信息。[Purpose/significance] It has significance to rank credible reviews for customer online shopping decision-making from massive and disordered shopping reviews.[Method/process] Firstly,after giving the definition of review credibility we extended knowledge graph model with weight and established knowledge graph with weight for describing product nature information model using weighted knowledge from the sources of evaluation of goods,and knowledge of domain experts.Secondly,from the product reviews we constructed review semantic networks by extracting knowledge reflecting product nature information and extracting product feature-opinion pair coincided with product nature information,and then calculated the value of review credibility according to review semantic network,and sorted on the values of review credibility.[Result/conclusion] Findings show that this model can rank online product reviews with credible information in e-shops,so that it can help potential customers acquire reference information.

关 键 词：带权知识图谱 商品本质信息 评论可信性 语义网 

分 类 号：F724.6[经济管理—产业经济] F274[文化科学—情报学] G353.1