What makes a review a reliable rating in recommender systems?
作者：
Highlights：
• Converting textual reviews to numeric ratings inherently encompasses uncertainty.
• Factors affecting the reliability of user review-to-rating conversion are studied.
• A confidence level is computed for ratings produced from user reviews.
• The confidence level is considered in the recommendation formulation process.
• Use of the confidence level increases prediction accuracy and user satisfaction.
摘要
•Converting textual reviews to numeric ratings inherently encompasses uncertainty.•Factors affecting the reliability of user review-to-rating conversion are studied.•A confidence level is computed for ratings produced from user reviews.•The confidence level is considered in the recommendation formulation process.•Use of the confidence level increases prediction accuracy and user satisfaction.
论文关键词：Recommender systems,Collaborative filtering,Rating prediction using textual information,Confidence level,Feature selection
论文评审过程：Received 12 October 2019, Revised 10 May 2020, Accepted 13 May 2020, Available online 5 June 2020, Version of Record 5 June 2020.
论文官网地址：https://doi.org/10.1016/j.ipm.2020.102304
