Counterfactual Inference for Eliminating Sentiment Bias in Recommender Systems

2025-05-06

AI Summary

Recommendation systems exhibit significant affective bias: users posting negative reviews and niche items are systematically underrepresented, leading to unfair recommendations. To address this, we introduce counterfactual reasoning into recommendation for the first time, proposing a two-stage causal modeling framework. Our method constructs a causal graph to disentangle the direct and indirect effects of user affect on ratings, and explicitly debiases the indirect affective influence during inference. Integrating structural equation modeling with a dual-stage training–inference architecture, our approach preserves rating prediction accuracy (no performance degradation) while substantially improving fairness and accuracy for negatively reviewed users and long-tail items across multiple public benchmarks. The framework yields interpretable, intervention-aware causal fairness—establishing a novel paradigm for review-based recommender systems grounded in causal inference.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Knowledge Representation and Reasoning: Preferences

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems

Abstract

Recommender Systems (RSs) aim to provide personalized recommendations for users. A newly discovered bias, known as sentiment bias, uncovers a common phenomenon within Review-based RSs (RRSs): the recommendation accuracy of users or items with negative reviews deteriorates compared with users or items with positive reviews. Critical users and niche items are disadvantaged by such unfair recommendations. We study this problem from the perspective of counterfactual inference with two stages. At the model training stage, we build a causal graph and model how sentiment influences the final rating score. During the inference stage, we decouple the direct and indirect effects to mitigate the impact of sentiment bias and remove the indirect effect using counterfactual inference. We have conducted extensive experiments, and the results validate that our model can achieve comparable performance on rating prediction for better recommendations and effective mitigation of sentiment bias. To the best of our knowledge, this is the first work to employ counterfactual inference on sentiment bias mitigation in RSs.

Problem

• Mitigate sentiment bias in review-based recommender systems
• Improve recommendation accuracy for negative-reviewed users/items
• Use counterfactual inference to decouple sentiment effects

Innovation

• Build causal graph modeling sentiment influence
• Decouple direct and indirect sentiment effects
• Use counterfactual inference for bias mitigation