通过可解释的神经元调控，让推荐系统更公平地曝光冷门内容。
From Insight to Intervention: Interpretable Neuron Steering for Controlling Popularity Bias in Recommender Systems
- 用稀疏自编码器识别出影响热门偏见的关键神经元。
- 调整这些神经元激活值后，冷门物品曝光率提升37%以上。
- 既能提升公平性，又保持推荐准确率，适合需要透明控制的场景。
流行度偏见是推荐系统中的普遍问题，少数热门物品占据主导，而大多数冷门物品被忽视，降低推荐质量并导致不公平曝光。现有方法虽部分缓解此问题，但缺乏操作透明性。本文提出一种后处理方法 PopSteer，利用稀疏自编码器（SAE）在保持模型行为一致的同时，实现神经元层面的可解释性。通过引入偏好热门或冷门物品的合成用户，基于激活模式识别出编码流行度信号的神经元，并通过调节其激活值来干预推荐结果。在三个公开数据集上使用序列推荐模型进行实验，结果表明，PopSteer 显著提升公平性，冷门物品曝光率平均提升超过37%，同时对准确性影响极小，且提供可解释的洞察与细粒度的公平性-准确性权衡控制。
原文摘要 · Abstract (English)
Popularity bias is a pervasive challenge in recommender systems, where a few popular items dominate attention while the majority of less popular items remain underexposed. This imbalance can reduce recommendation quality and lead to unfair item exposure. Although existing mitigation methods address this issue to some extent, they often lack transparency in how they operate. In this paper, we propose a post-hoc approach, PopSteer, that leverages a Sparse Autoencoder (SAE) to both interpret and mitigate popularity bias in recommendation models. The SAE is trained to replicate a trained model's behavior while enabling neuron-level interpretability. By introducing synthetic users with strong preferences for either popular or unpopular items, we identify neurons encoding popularity signals through their activation patterns. We then steer recommendations by adjusting the activations of the most biased neurons. Experiments on three public datasets with a sequential recommendation model demonstrate that PopSteer significantly enhances fairness with minimal impact on accuracy, while providing interpretable insights and fine-grained control over the fairness-accuracy trade-off.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。