# Contrastive Learning for Debiased Candidate Generation in Large-Scale Recommender Systems

arXiv: https://arxiv.org/abs/2005.12964 · HTML: https://mlsanity.com/paper/2005.12964/

## Key facts

- Cited by 38 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 May 2020.

## Built on by (most cited first, 30 of 38)

- [BadCLIP: Dual-Embedding Guided Backdoor Attack on Multimodal Contrastive Learning](https://mlsanity.com/paper/2311.12075/) (2023)
- [Enhancing Activity Prediction Models in Drug Discovery with the Ability to Understand Human Language](https://mlsanity.com/paper/2303.03363/) (2023)
- [Personalized Showcases: Generating Multi-Modal Explanations for Recommendations](https://mlsanity.com/paper/2207.00422/) (2022)
- [A Systematic Review and Replicability Study of BERT4Rec for Sequential Recommendation](https://mlsanity.com/paper/2207.07483/) (2022)
- [Adap-τ: Adaptively Modulating Embedding Magnitude for Recommendation](https://mlsanity.com/paper/2302.04775/) (2023)
- [Modeling Techniques for Machine Learning Fairness: A Survey](https://mlsanity.com/paper/2111.03015/) (2021)
- [A Survey on Fairness-aware Recommender Systems](https://mlsanity.com/paper/2306.00403/) (2023)
- [Revisiting Neural Retrieval on Accelerators](https://mlsanity.com/paper/2306.04039/) (2023)
- [Recent Developments in Recommender Systems: A Survey](https://mlsanity.com/paper/2306.12680/) (2023)
- [RecDCL: Dual Contrastive Learning for Recommendation](https://mlsanity.com/paper/2401.15635/) (2024)
- [Negating Negatives: Alignment with Human Negative Samples via Distributional Dispreference Optimization](https://mlsanity.com/paper/2403.03419/) (2024)
- [Toward a Better Understanding of Loss Functions for Collaborative Filtering](https://mlsanity.com/paper/2308.06091/) (2023)
- [Scenario-Adaptive and Self-Supervised Model for Multi-Scenario Personalized Recommendation](https://mlsanity.com/paper/2208.11457/) (2022)
- [Imitate the Good and Avoid the Bad: An Incremental Approach to Safe Reinforcement Learning](https://mlsanity.com/paper/2312.10385/) (2023)
- [Revisiting Recommendation Loss Functions through Contrastive Learning (Technical Report)](https://mlsanity.com/paper/2312.08520/) (2023)
- [Cold-start Bundle Recommendation via Popularity-based Coalescence and Curriculum Heating](https://mlsanity.com/paper/2310.03813/) (2023)
- [Generating Negative Samples for Sequential Recommendation](https://mlsanity.com/paper/2208.03645/) (2022)
- [eSASRec: Enhancing Transformer-based Recommendations in a Modular Fashion](https://mlsanity.com/paper/2508.06450/) (2025)
- [UniMatch: A Unified User-Item Matching Framework for the Multi-purpose Merchant Marketing](https://mlsanity.com/paper/2307.09989/) (2023)
- [3D pride without 2D prejudice: Bias-controlled multi-level generative models for structure-based ligand design](https://mlsanity.com/paper/2204.10663/) (2022)
- [Using Interventions to Improve Out-of-Distribution Generalization of Text-Matching Recommendation Systems](https://mlsanity.com/paper/2210.10636/) (2022)
- [Towards Assumption-free Bias Mitigation](https://mlsanity.com/paper/2307.04105/) (2023)
- [Boosting Fair Classifier Generalization through Adaptive Priority Reweighing](https://mlsanity.com/paper/2309.08375/) (2023)
- [Towards Better Modeling with Missing Data: A Contrastive Learning-based Visual Analytics Perspective](https://mlsanity.com/paper/2309.09744/) (2023)
- [Leveraging Negative Signals with Self-Attention for Sequential Music Recommendation](https://mlsanity.com/paper/2309.11623/) (2023)
- [GPS: Graph Contrastive Learning via Multi-scale Augmented Views from Adversarial Pooling](https://mlsanity.com/paper/2401.16011/) (2024)
- [Cross-domain User Preference Learning for Cold-start Recommendation](https://mlsanity.com/paper/2112.03667/) (2021)
- [Improving Multi-Interest Network with Stable Learning](https://mlsanity.com/paper/2207.07910/) (2022)
- [(Debiased) Contrastive Learning Loss for Recommendation (Technical Report)](https://mlsanity.com/paper/2312.08517/) (2023)
- [Bilateral Unsymmetrical Graph Contrastive Learning for Recommendation](https://mlsanity.com/paper/2403.15075/) (2024)

## Abstract (arXiv)

Deep candidate generation (DCG) that narrows down the collection of relevant items from billions to hundreds via representation learning has become prevalent in industrial recommender systems. Standard approaches approximate maximum likelihood estimation (MLE) through sampling for better scalability and address the problem of DCG in a way similar to language modeling. However, live recommender systems face severe exposure bias and have a vocabulary several orders of magnitude larger than that of natural language, implying that MLE will preserve and even exacerbate the exposure bias in the long run in order to faithfully fit the observed samples. In this paper, we theoretically prove that a popular choice of contrastive loss is equivalent to reducing the exposure bias via inverse propensity weighting, which provides a new perspective for understanding the effectiveness of contrastive learning. Based on the theoretical discovery, we design CLRec, a contrastive learning method to improve DCG in terms of fairness, effectiveness and efficiency in recommender systems with extremely large candidate size. We further improve upon CLRec and propose Multi-CLRec, for accurate multi-intention aware bias reduction. Our methods have been successfully deployed in Taobao, where at least four-month online A/B tests and offline analyses demonstrate its substantial improvements, including a dramatic reduction in the Matthew effect.