Counterfactual Reasoning for Explainable Recommendation in Path-Based Models
The author argues that attention-based explainability in recommendation models lacks stability and effectiveness, proposing a novel framework based on counterfactual reasoning to enhance path-based explanations.