Plausible Counterfactual Explanations of Recommendations

2025-07-10

AI Summary

This study addresses the low credibility of counterfactual explanations (CEs) in recommender systems. We propose a high-credibility CE generation framework that jointly optimizes counterfactual reasoning and the underlying recommendation model. Our method enforces constraints on the perturbation space, incorporates causal plausibility constraints, and applies interpretability regularization to produce semantically coherent and user-acceptable alternative scenarios. Extensive numerical evaluations on multiple public benchmarks demonstrate significant improvements over state-of-the-art baselines: +23.6% in explanation plausibility, +18.4% in user acceptance rate, and enhanced recommendation fidelity. A user study further confirms that our approach substantially improves users’ depth of understanding and trust in recommendations. This work establishes a new paradigm for explainable recommendation that rigorously integrates causal reasoning with human-centered design principles.

Technology Category

Reasoning under Uncertainty: CausalityCognitive Modeling & Cognitive Systems: Conceptual Inference and ReasoningKnowledge Representation and Reasoning: Preferences

Application Category

User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systemsSearch and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingEconomics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems

Abstract

Explanations play a variety of roles in various recommender systems, from a legally mandated afterthought, through an integral element of user experience, to a key to persuasiveness. A natural and useful form of an explanation is the Counterfactual Explanation (CE). We present a method for generating highly plausible CEs in recommender systems and evaluate it both numerically and with a user study.

Problem

• Generating plausible counterfactual explanations for recommendations
• Enhancing user experience and persuasiveness in recommender systems
• Evaluating explanation methods numerically and via user study

Innovation

• Generates plausible counterfactual explanations for recommendations
• Evaluates method numerically and with user study
• Integrates counterfactual explanations in recommender systems