基于知识采纳模型的在线评论有用性识别——以大众点评网为例
发布时间:2022-11-12 13:16:55
2017年6月 June 2017 情报探索 Information Researeh 第6期(总236期) No.6(Serial No.236) 基于知识采纳模型的在线评论有用性识别 ——以大众点评网为例 唐艺楠徐德华 200092) (同济大学经济与管理学院 上海摘要:[目的/意义]旨在自动识别高质量的在线评论 [方法/过程]以大众点评网某商家2398条评论为样本,提取其评论文 本特征和评论者特征,采用监督学习的方法进行评论质量识别。【结果,结论]5种质量分类算法中,GradientBoosting模型和Ad. aBoosting模型效果最好。这2种模型中,对分类结果贡献度最高的是评论词语总数,评论文本广度对评论有用性有正向影响:相 对于消极情感值,积极情感值有更大的贡献度;评论者特征中,评论者的贡献值对评论有用性影响最大。 关键词:在线餐饮评论;有用性;评论文本特征;评论者特征;文本分类 中图分类号:F713.365.2 文献标识码:A doklO.3969 ̄.issn.1005—8095.2017.06.002 Knowledge Adoption Model—-based Online Reviews Helpfulness Identiicatifon: Case Study of Dianping Website Tang Yinan Xu Dehua (School of Economics and Management of Tongji University,Shanghai 200092) Abstract:[Purpose/sigulficance]The paper is to automatically identify high quality online reviews.[Method/process]The paper takes a memhant’S 2398 reviews on Dianping website as sample,extracts their text features and reviewer features,and adopts super- vised learning method to do reviews helpfulness identiifcation.[Result/conclusion]Among 5 quality classiifcation algoirthms,Gradient— Boosting model and AdaBoosting model work the best.These two models show that the highest contribution to classiifcation results is total number of commentary words;the width of review text has posit