Causality-aware Graph Aggregation Weight Estimator for Popularity Debiasing in Top-K Recommendation

2025-10-06

AI Summary

Graph neural recommendation systems suffer from popularity bias and echo effects due to over-reliance on popular items during neighbor aggregation. Existing debiasing methods lack theoretical guarantees and suffer from imbalanced training-debiasing dynamics. This paper identifies neighbor aggregation as a form of backdoor adjustment in causal inference and proposes CAGED, a Causal-Aware Graph Debiasing framework. CAGED employs variational inference to optimize the evidence lower bound of interaction likelihood, enabling unbiased estimation of neighbor weights; it further incorporates a momentum-based update mechanism to enhance robustness against bias propagation. Extensive experiments on three public benchmarks demonstrate that CAGED significantly mitigates popularity bias, consistently outperforming state-of-the-art debiasing methods in Recall@K and NDCG@K while preserving both recommendation accuracy and fairness.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• 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

Abstract

Graph-based recommender systems leverage neighborhood aggregation to generate node representations, which is highly sensitive to popularity bias, resulting in an echo effect during information propagation. Existing graph-based debiasing solutions refine the aggregation process with attempts such as edge reconstruction or weight adjustment. However, these methods remain inadequate in fully alleviating popularity bias. Specifically, this is because 1) they provide no insights into graph aggregation rationality, thus lacking an optimality guarantee; 2) they fail to well balance the training and debiasing process, which undermines the effectiveness. In this paper, we propose a novel approach to mitigate popularity bias through rational modeling of the graph aggregation process. We reveal that graph aggregation is a special form of backdoor adjustment in causal inference, where the aggregation weight corresponds to the historical interaction likelihood distribution. Based on this insight, we devise an encoder-decoder architecture, namely Causality-aware Graph Aggregation Weight Estimator for Debiasing (CAGED), to approximate the unbiased aggregation weight by optimizing the evidence lower bound of the interaction likelihood. In order to enhance the debiasing effectiveness during early training stages, we further design a momentum update strategy that incrementally refines the aggregation weight matrix. Extensive experiments on three datasets demonstrate that CAGED outperforms existing graph-based debiasing methods. Our implementation is available at https://github.com/QueYork/CAGED.

Problem

• Estimating unbiased graph aggregation weights for recommendations
• Mitigating popularity bias in graph-based recommender systems
• Balancing training effectiveness with debiasing in graph aggregation

Innovation

• Models graph aggregation as causal backdoor adjustment
• Estimates unbiased weights via encoder-decoder architecture
• Uses momentum update strategy for progressive debiasing