论文部分内容阅读
【目的】从大量在线商品评论中筛选出可信的评论辅助消费者制定购买决策。【方法】提出一种基于大数据思维的主流特征观点对的概念,依据特征观点对在不同用户评论中的认可程度,建立评论可信性排序模型。【结果】淘宝、天猫和京东平台的商品评论的主流特征观点对是稳定的;与已有模型相比,使用本文模型排序过的用户评论包含的产品特征范围更广,评论有用性提升7.5%,更能够反映评论的真实情况。【局限】仅从评论包含的特征观点对考虑评论可信性,而未考虑评论的具体语义情况。【结论】包含主流特征观点对数量越多的评论,其可信度则越大。
[Purpose] To screen credible comments from a large number of online product reviews to assist consumers in making purchase decisions. 【Method】 A concept of mainstream feature point based on big data thinking was proposed. Based on the viewpoints of features, the author approved the degree of recognition in different user comments, and established a trustworthiness ranking model. [Results] The mainstream features of the product reviews in Taobao, Tmall and Jingdong platforms are stable. Compared with the existing models, the user reviews sorted by this model contain a wider range of product features, and the usefulness of comments is improved by 7.5 %, More able to reflect the real situation of the commentary. Limitations Consider the credibility of comments only from the perspective of features included in the comments, without regard to the specific semantic situation of the comments. CONCLUSION: The larger the number of comments on the mainstream features, the more credible.