BPL: Bias-Adaptive Preference Distillation Learning For Recommender System
- 9 October 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (1) , 110-122
Abstract
Recommender systems suffer from biases that cause the collected feedback to incompletely reveal user preference. While debiasing learning has been extensively studied, they mostly focused on the specialized (called counterfactual) test environment simulated by random exposure of items, significantly degrading accuracy in the typical (called factual) test environment based on actual user-item interactions. In fact, each test environment highlights the benefit of a different aspect: the counterfactual test emphasizes user satisfaction in the long-terms, while the factual test focuses on predicting subsequent user behaviors on platforms. Therefore, it is desirable to have a model that performs well on both tests rather than only one. In this work, we introduce a new learning framework, called Bias-adaptive Preference distillation Learning (BPL), to gradually uncover user preferences with dual distillation strategies. These distillation strategies are designed to drive high performance in both factual and counterfactual test environments. Employing a specialized form of teacher-student distillation from a biased model, BPL retains accurate preference knowledge aligned with the collected feedback, leading to high performance in the factual test. Furthermore, through self-distillation with reliability filtering, BPL iteratively refines its knowledge throughout the training process. This enables the model to produce more accurate predictions across a broader range of user-item combinations, thereby improving performance in the counterfactual test. Comprehensive experiments validate the effectiveness of BPL in both factual and counterfactual tests.Keywords
All Related Versions
Funding Information
- Microsoft Research Asia
- IITP (2022-00155958)
- Basic Science Research Program
- Ministry of Education (NRF-2021R1A6A1A03045425)
- ICT Creative Consilience program
- IITP
- Ministry of Science and ICT, South Korea (IITP-2025-RS-2020-II201819)
- IITP
- Ministry of Science and ICT, South Korea (RS-2024-00457882)
- 2025 High-Performance Computing Support Program
This publication has 44 references indexed in Scilit:
- Meta doubly robust: Debiasing CVR prediction via meta-learning with a small amount of unbiased dataKnowledge-Based Systems, 2024
- Uncovering the Propensity Identification Problem in Debiased Recommendations2024
- Bilateral Improvement in Local Personalization and Global Generalization in Federated LearningIEEE Internet of Things Journal, 2024
- Popularity Bias is not Always Evil: Disentangling Benign and Harmful Bias for RecommendationIEEE Transactions on Knowledge and Data Engineering, 2022
- KDCRec: Knowledge Distillation for Counterfactual Recommendation via Uniform DataIEEE Transactions on Knowledge and Data Engineering, 2022
- Undoing the Damage of Label Shift for Cross-domain Semantic Segmentation2022
- Re-energizing Domain Discriminator with Sample Relabeling for Adversarial Domain Adaptation2021
- Classes Matter: A Fine-Grained Adversarial Approach to Cross-Domain Semantic Segmentation2020
- A theory of learning from different domainsMachine Learning, 2009
- Matrix Factorization Techniques for Recommender SystemsComputer, 2009