# PEARL: Unbiased Percentile Estimation via Contrastive Learning for Industrial-Scale Livestream Recommendation

arXiv: https://arxiv.org/abs/2605.21752 · HTML: https://mlsanity.com/paper/2605.21752/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 15 of its 23 references that are in the map.
- First posted on arXiv 20 May 2026.

## Builds on (most cited first, 15 of 15)

- [Variational Autoencoders for Collaborative Filtering](https://mlsanity.com/paper/1802.05814/) (2018)
- [Recommendations as Treatments: Debiasing Learning and Evaluation](https://mlsanity.com/paper/1602.05352/) (2016)
- [Neural Attentive Session-based Recommendation](https://mlsanity.com/paper/1711.04725/) (2017)
- [Unbiased Learning-to-Rank with Biased Feedback](https://mlsanity.com/paper/1608.04468/) (2016)
- [Learning Disentangled Representations for Recommendation](https://mlsanity.com/paper/1910.14238/) (2019)
- [Wukong: Towards a Scaling Law for Large-Scale Recommendation](https://mlsanity.com/paper/2403.02545/) (2024)
- [The Unfairness of Popularity Bias in Recommendation](https://mlsanity.com/paper/1907.13286/) (2019)
- [AutoDebias: Learning to Debias for Recommendation](https://mlsanity.com/paper/2105.04170/) (2021)
- [Deconfounding Duration Bias in Watch-time Prediction for Video Recommendation](https://mlsanity.com/paper/2206.06003/) (2022)
- [The Deconfounded Recommender: A Causal Inference Approach to Recommendation](https://mlsanity.com/paper/1808.06581/) (2018)
- [DVR: Micro-Video Recommendation Optimizing Watch-Time-Gain under Duration Bias](https://mlsanity.com/paper/2208.05190/) (2022)
- [Conditional Quantile Estimation for Uncertain Watch Time in Short-Video Recommendation](https://mlsanity.com/paper/2407.12223/) (2024)
- [Relative Advantage Debiasing for Watch-Time Prediction in Short-Video Recommendation](https://mlsanity.com/paper/2508.11086/) (2025)
- [Zenith: Scaling up Ranking Models for Billion-scale Livestreaming Recommendation](https://mlsanity.com/paper/2601.21285/) (2026)
- [ContentCTR: Frame-level Live Streaming Click-Through Rate Prediction with Multimodal Transformer](https://mlsanity.com/paper/2306.14392/) (2023)

## Abstract (arXiv)

Recommender systems trained on user interaction data are susceptible to behavioral intensity imbalance--a systematic distortion arising from heterogeneous engagement patterns across users. This imbalance skews feedback signals such that observed interactions no longer faithfully reflect true preferences, causing models to disproportionately amplify signals from highly active users while underrepresenting others, which ultimately degrades recommendation quality and robustness at scale. To address this issue, we propose a nonparametric contrastive percentile approximation framework, PEARL, that models relative preference signals instead of absolute engagement magnitudes. Building upon relative advantage debiasing, PEARL leverages real contrastive interaction samples to approximate percentile relationships directly, without relying on auxiliary distribution estimation models. We provide theoretical justification demonstrating that such pairwise comparisons yield unbiased estimates of percentile-based preference signals. For broader applicability, we introduce a prediction-based bootstrapping mechanism for percentile smoothing to handle sparse and discrete feedback, alongside a generalized value-weighted formulation and a co-training strategy to enhance both modeling flexibility and representation learning. Extensive offline experiments demonstrate that PEARL effectively mitigates behavioral bias and consistently improves recommendation performance across multiple ranking targets. Deployed in a production livestream platform with a combined user base of billions, online A/B testing confirms substantial real-world gains: +2.10% Watch Duration, +0.80% Consumption Amount, +1.49% Interaction Rate, and -6.91% Report Rate.