摘要
【目的】对在线产品评论垃圾识别的研究工作进行梳理,总结研究现状,明确发展方向。【文献范围】从CNKI及Google Scholar中以"评论垃圾"、"review/opinion spam"等为检索词筛选获得国内外近50篇相关文献。【方法】采用文献分析法。界定产品评论垃圾的概念,明晰评论垃圾所属的研究范畴;总结产品评论垃圾识别研究中的关键问题及研究进展。【结果】产品评论垃圾指故意过分吹捧或贬低某种产品的不真实评论以及不包含任何有益成分的非相关在线网络评论。研究中存在虚假评论标注集难获取的问题,强调评论人行为特征的分析,提出融合评论人特征来解决评论垃圾识别的研究思路。【局限】应结合产品评论垃圾的识别对用户评论可信度进行更深入的分析。【结论】评论垃圾识别是评论可信性研究的应用体现。辩识评论内容的真伪要充分挖掘评论内容、评论人等多个维度的识别特征。同时,考虑到众多特征的相互独立性,应挖掘有显著影响作用的特征因素。
[Objective] The paper cards existing study about online product review spam, summarizes research status, and puts forward future research direction. [Coverage] Almost 50 papers at home and abroad are searched via review/opinion spam as keywords from CNKI and Google scholar. [Methods] By literature analysis, the concept of product review spam is defined. The research area that review spam study belongs to is specified and key issues and challenges are presented in the paper. [Results] Product review spam refers to the untruthful reviews written for the purpose of inflating or damaging given products excessively or low quality reviews not being able to provide any help to customers. Due to the lack of reliable ground truth label of fake/no-fake review data, the analysis for reviewers＇ behavior is highlighted since it can be employed to solve the problem of fake review identification effectively if being combined with the features of review contents. [Limitations] Further study should be conducted on the creditability analysis for product review spam combined with fake review identification. [Conclusions] Product review spam detection is a kind of application studies corresponding to review creditability. Not only the review content specific features but also reviewers corresponding features should be fully explored for fake reviews detection. Moreover, the features with significant impact on fake review identification need to be highlighted specifically with the consideration of independence of feature variables.
出处
《现代图书情报技术》
CSSCI
北大核心
2014年第2期63-71,共9页
New Technology of Library and Information Service
基金
广东省哲学社会科学"十二五"规划2013年度项目"基于情境和用户感知的知识推荐机制研究"(项目编号:CD13CTS01)的研究成果之一
关键词
评论垃圾
评论可信度
评论有用性
Review spam Credibility of review Helpfulness of review
