Why This, Not That? Mining User Profiles for Pair-wise Counterfactuals

2026-08-21

AI Summary

研究通过反事实学习方法，解决了推荐系统中为何一个项目比另一个排名更高的解释问题，基于用户档案挖掘相关因素。

Abstract

The topic of explanation in recommender systems has seen steady research attention since the earliest days of the field. With some exceptions, this work has focused on the explanation of single items in a recommendation list and, especially recently, has emphasized approaches that are decoupled from the logic of the recommendation algorithm itself. Based on findings in the psychology of interpersonal communication, we propose a new task, pairwise interpretation of item rankings, asking the comparative question ``Why is item A ranked higher than item B?''. An effective solution to this task, we argue, is inherently grounded in the operation of the recommendation algorithm. We propose a class of techniques based on counterfactual learning to uncover the items in a user's profile that have contributed to the relative ranking of items. Using multiple datasets, we show that it is possible to identify such items as potential basis for comparative explanation.

Problem

• pairwise interpretation
• item rankings
• recommendation algorithm
• counterfactual learning
• user profiles

Innovation

• pairwise interpretation
• counterfactual learning
• item rankings