可解释信息推荐研究综述

A Review of Explainable Information Recommendation

可解释信息推荐能有效提高推荐结果的可靠性,提升用户信任度和满意度,是未来研究的重要方向之一。了解可解释信息推荐的研究进展,能为可解释推荐研究提供新的思路与借鉴,促进可解释推荐研究的发展。通过对国内外可解释信息推荐相关文献的梳理与分析,深入剖析可解释信息推荐算法与解释机理、推荐解释的媒介与呈现,以及推荐解释评价3个方面的研究现状与不足,尝试对可解释智能推荐系统所关注的“三个W(who,what,why)问题”进行回答,并提出了可解释信息推荐未来的研究方向与课题。

Explainable information recommendation can effectively improve the reliability of recommendation results and user’s trust and satisfaction,which will be an important direction of future research.Reviewing the situation and progress of research in this area can provide ideas and reference for further investigation,and promote the development of explainable recommender.By searching and investigating the literature related to explainable information recommendation at home and abroad,we deeply analyzed the status and shortcomings of research in three aspects:interpretation mechanism of explainable recommendation model,media and presentation of recommendation interpretation,and recommendation interpretation evaluation.Furthermore,we responded to the“three W(who,what,why)questions”in explainable intelligent recommender system,proposed future research directions and topics of explainable information recommendation.