CRAB: Codebook Rebalancing for Bias Mitigation in Generative Recommendation

TL;DR

This paper introduces CRAB, a post-hoc debiasing method for generative recommendation models that reduces popularity bias by rebalancing semantic token representations and improving their semantic consistency.

Contribution

CRAB is a novel post-hoc strategy that rebalances token frequency and enhances semantic structure to mitigate popularity bias in GeneRec models.

Findings

CRAB significantly reduces popularity bias in recommendation tasks.

CRAB improves recommendation accuracy on real-world datasets.

Rebalancing tokens enhances semantic consistency and informativeness.

Abstract

Generative recommendation (GeneRec) has introduced a new paradigm that represents items as discrete semantic tokens and predicts items in a generative manner. Despite its strong performance across multiple recommendation tasks, existing GeneRec approaches still suffer from severe popularity bias and may even exacerbate it. In this work, we conduct a comprehensive empirical analysis to uncover the root causes of this phenomenon, yielding two core insights: 1) imbalanced tokenization inherits and can further amplify popularity bias from historical item interactions; 2) current training procedures disproportionately favor popular tokens while neglecting semantic relationships among tokens, thereby intensifying popularity bias. Building on these insights, we propose CRAB, a post-hoc debiasing strategy for GeneRec that alleviates popularity bias by mitigating frequency imbalance among semantic tokens. Specifically, given a well-trained model, we first rebalance the codebook by splitting over-popular tokens while preserving their hierarchical semantic structure. Based on the adjusted codebook, we further introduce a tree-structured regularizer to enhance semantic consistency, encouraging more informative representations for unpopular tokens during training. Experiments on real-world datasets demonstrate that CRAB significantly improves recommendation performance by effectively alleviating popularity bias.