mlsanityData to 14 Oct 2026

Ranking with Popularity Bias: User Welfare under Self-Amplification Dynamics

arXiv 2305.18333 ↗First posted 24 May 2023cs.IR cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

While popularity bias is recognized to play a crucial role in recommmender (and other ranking-based) systems, detailed analysis of its impact on collective user welfare has largely been lacking. We propose and theoretically analyze a general mechanism, rooted in many of the models proposed in the literature, by which item popularity, item quality, and position bias jointly impact user choice. We focus on a standard setting in which user utility is largely driven by item quality, and a recommender attempts to estimate it given user behavior. Formulating the problem as a non-stationary contextual bandit, we study the ability of a recommender policy to maximize user welfare under this model. We highlight the importance of exploration, not to eliminate popularity bias, but to mitigate its negative impact on welfare. We first show that naive popularity-biased recommenders induce linear regret by conflating item quality and popularity. More generally, we show that, even in linear settings, identifiability of item quality may not be possible due to the confounding effects of popularity bias. However, under sufficient variability assumptions, we develop an efficient optimistic algorithm and prove efficient regret guarantees w.r.t. user welfare. We complement our analysis with several simulation studies, which demonstrate the negative impact of popularity bias on the performance of several natural recommender policies.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 30 of its 85 references that are in the map. First posted on arXiv 24 May 2023.

Builds on

30 of 85 references.

References · 30

  1. 2014Sketching as a Tool for Numerical Linear Algebracited by 288
  2. 2018Fairness of Exposure in Rankingscited by 190
  3. 2016Unbiased Learning-to-Rank with Biased Feedbackcited by 157
  4. 2020Improved Optimistic Algorithms for Logistic Banditscited by 118
  5. 2021Causal Intervention for Leveraging Popularity Bias in Recommendationcited by 116
  6. 2017Mostly Exploration-Free Algorithms for Contextual Banditscited by 85
  7. 2019Managing Popularity Bias in Recommender Systems with Personalized Re-rankingcited by 82
  8. 2016Explore First, Exploit Next: The True Shape of Regret in Bandit Problemscited by 79
  9. 2015Cascading Bandits: Learning to Rank in the Cascade Modelcited by 77
  10. 2019The Unfairness of Popularity Bias in Recommendationcited by 70
  11. 2020Controlling Fairness and Bias in Dynamic Learning-to-Rankcited by 68
  12. 2020Instance-Wise Minimax-Optimal Algorithms for Logistic Banditscited by 57
18 more
  1. 2016Cascading Bandits for Large-Scale Recommendation Problemscited by 45
  2. 2021User-centered Evaluation of Popularity Bias in Recommender Systemscited by 38
  3. 2020The Connection Between Popularity Bias, Calibration, and Fairness in Recommendationcited by 35
  4. 2021Leveraging Good Representations in Linear Contextual Banditscited by 29
  5. 2019OSOM: A simultaneously optimal algorithm for multi-armed and linear contextual banditscited by 28
  6. 2017Online Learning to Rank in Stochastic Click Modelscited by 26
  7. 2020Multi-sided Exposure Bias in Recommendationcited by 23
  8. 2021Popularity Bias Is Not Always Evil: Disentangling Benign and Harmful Bias for Recommendationcited by 23
  9. 2020Bandits with Partially Observable Confounded Datacited by 22
  10. 2018Semiparametric Contextual Banditscited by 19
  11. 2018Thompson Sampling Algorithms for Cascading Banditscited by 13
  12. 2021UCB-based Algorithms for Multinomial Logistic Regression Banditscited by 12
  13. 2023Reinforcement Learning with History-Dependent Dynamic Contextscited by 11
  14. 2020Popularity Bias in Recommendation: A Multi-stakeholder Perspectivecited by 6
  15. 2022Reinforcement Learning with a Terminatorcited by 4
  16. 2020Disentangling User Interest and Conformity for Recommendation with Causal Embeddingcited by 2
  17. 2022Evolution of Popularity Bias: Empirical Study and Debiasingcited by 2
  18. 2023Overcoming Prior Misspecification in Online Learning to Rankcited by 1

Built on by

Not cited yet.