Gender-aware fairness in iterative recommender systems: A simulation study on popularity bias

Öz

This study examines how gender-based disparities emerge in recommender systems through feedback loops. While fairness has been studied in static settings, little is known about how repeated user-system interactions impact different demographic groups over time. To address this, we utilize a dynamic simulation framework using synthetic interactions and ten feedback iterations. Based on the MovieLens-1M dataset, users are grouped by gender and evaluated using metrics such as calibration, diversity, and long-tail exposure. Results show that female users consistently receive less favorable outcomes, with popularity bias measures (GAP, MRMC) indicating a growing disadvantage over time. Diversity and novelty scores also decline more sharply for women. These findings suggest that feedback loops can reinforce existing inequalities in recommender systems. The employed framework provides a valuable tool for analyzing the evolution of fairness across iterations and highlights the need for gender-sensitive algorithms that maintain fairness over time.

Anahtar Kelimeler

• Gender fairness
• recommender systems
• feedback loop
• popularity bias
• demographic disparity

Etik Beyan

The authors declare that they have no conflict of interest.

Teşekkür

This work is based on the Master’s thesis of Yildiz Zoralioglu and builds upon the research conducted as part of her graduate studies.

İteratif öneri sistemlerinde cinsiyete-duyarlı adalet: Popülerlik yanlılığı üzerine bir simülasyon çalışması

Öz

Bu çalışma, öneri sistemlerinde geri besleme döngüleri yoluyla cinsiyete dayalı eşitsizliklerin nasıl ortaya çıktığını incelemektedir. Adalet konusu durağan ortamlarda araştırılmış olsa da yinelenen kullanıcı-sistem etkileşimlerinin zaman içinde farklı demografik grupları nasıl etkilediği hakkında çok az bilgi bulunmaktadır. Bu durumu ele almak için, sentetik etkileşimler ve on geri besleme iterasyonu içeren dinamik bir simülasyon çerçevesi kullanılmıştır. MovieLens-1M veri kümesine dayalı olarak kullanıcılar cinsiyete göre gruplanmış ve kalibrasyon, çeşitlilik ve uzun kuyruk içeriklere erişim gibi metriklerle değerlendirilmiştir. Sonuçlar, kadın kullanıcıların sistemden sürekli olarak daha olumsuz sonuçlar aldığını göstermekte; GAP ve MRMC gibi popülerlik yanlılığı metrikleri ise zamanla artan bir dezavantajı ortaya koymaktadır. Ayrıca, kadın kullanıcılar için çeşitlilik ve yenilik skorlarının daha keskin bir şekilde düştüğü gözlemlenmiştir. Bu bulgular, geri besleme döngülerinin öneri sistemlerinde mevcut eşitsizlikleri pekiştirebileceğini ortaya koymakta ve zaman içinde adaleti koruyacak cinsiyete duyarlı algoritmalara duyulan ihtiyacı vurgulamaktadır.

Anahtar Kelimeler

• Cinsiyet adaleti
• öneri sistemleri
• geri besleme döngüsü
• popülerlik yanlılığı
• demografik eşitsizlik

Kaynakça

1. G. Adomavicius and A. Tuzhilin, Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions. IEEE Transactions on Knowledge and Data Engineering, 17 (6), 734–749, 2005. https://doi.org/10.1109/TKDE.2005.99.
2.    F. Ricci, L. Rokach and B. Shapira, Recommender Systems Handbook. Springer, 2015.
3.    X. Su and T. M. Khoshgoftaar, A survey of collaborative filtering techniques. Advances in Artificial Intelligence, 2009, Article ID 421425, 1–19, 2009. https://doi.org/10.1155/2009/421425.
4.    Y. Koren, R. Bell and C. Volinsky, Matrix factorization techniques for recommender systems. Computer, 42 (8), 30–37, 2009. https://doi.org/10.1109/MC.2009.263.
5.    G. Linden, B. Smith and J. York, Amazon.com recommendations: Item-to-item collaborative filtering. IEEE Internet Computing, 7 (1), 76–80, 2003. https://doi.org/10.1109/MIC.2003.1167344.
6.    G. Adomavicius and Y. Kwon, Improving aggregate recommendation diversity using ranking-based techniques. IEEE Transactions on Knowledge and Data Engineering, 24 (5), 896–911, 2011. https://doi.org/10.1109/TKDE.2011.15.
7.    H. Abdollahpouri, M. Mansoury, R. Burke and B. Mobasher, The unfairness of popularity bias in recommendation. Proceedings of the Workshop on Fairness, Accountability, and Transparency in Machine Learning (FAT/ML), 1–5, Long Beach, CA, USA, 2019.
8.    L. Boratto, G. Fenu and M. Marras, Connecting user and item perspectives in popularity debiasing for collaborative recommendation. Information Processing & Management, 58 (1), 102387, 2021. https://doi.org/10.1016/j.ipm.2020.102387.

9.    H. Abdollahpouri, M. Mansoury, R. Burke, B. Mobasher and E. Malthouse, User-centered evaluation of popularity bias in recommender systems. Proceedings of the 29th ACM Conference on User Modeling, Adaptation and Personalization (UMAP '21), 119–128, Utrecht, Netherlands, 2021. https://doi.org/10.1145/3450613.3456821.
10. R. Sinha and K. Swearingen, Comparing recommendations made by online systems and by friends. DELOS-NSF Workshop on Personalization and Recommender Systems in Digital Libraries, 1–10, Dublin, Ireland, 2001.
11. H. Abdollahpouri, R. Burke and B. Mobasher, Managing popularity bias in recommender systems with personalized re-ranking. Proceedings of the 32nd International Florida Artificial Intelligence Research Society Conference (FLAIRS), 413–418, Sarasota, FL, USA, 2019.
12. D. Turnbull, S. McQuillan, S. Zhang and D. Morrison, Exploring popularity bias in music recommendation models and commercial systems. arXiv preprint, arXiv:2208.09517, 2022. https://doi.org/10.48550/arXiv.2208.09517.
13. H. Abdollahpouri, M. Mansoury, R. Burke and B. Mobasher, The connection between popularity bias, calibration, and fairness in recommendation. Proceedings of the 14th ACM Conference on Recommender Systems (RecSys '20), 726–731, Virtual Event, Brazil, 2020. https://doi.org/10.1145/3383313.3418487.
14. R. Mehrotra, J. McInerney, H. Bouchard, M. Lalmas and F. Diaz, Towards a fair marketplace: Counterfactual evaluation of the fairness of exposure in recommender systems. Proceedings of the 27th ACM International Conference on Information and Knowledge Management (CIKM '18), 2243–2251, 2018. https://doi.org/10.1145/3269206.3272027.
15. D. Kowald, M. Schedl and E. Lex, The unfairness of popularity bias in music recommendation: A reproducibility study. Advances in Information Retrieval. ECIR 2020. Lecture Notes in Computer Science, 12036, 35–42, 2020. https://doi.org/10.1007/978-3-030-45442-5_5.
16. M. D. Ekstrand, M. Tian, I. M. Azpiazu, J. D. Ekstrand, O. Anuyah, D. McNeill and M. S. Pera, All the cool kids, how do they fit in? Popularity and demographic biases in recommender evaluation and effectiveness. Proceedings of the 1st Conference on Fairness, Accountability and Transparency (FAT '18), 172–186, New York, NY, USA, 2018.
17. M. Gulsoy, E. Yalcin and A. Bilge, Robustness of privacy-preserving collaborative recommenders against popularity bias problem. PeerJ Computer Science, 9, e1438, 2023. https://doi.org/10.7717/peerj-cs.1438.
18. B. Ferwerda, M. Schedl and M. Tkalčič, Personality and taxonomy preferences, gender, and age in music recommender systems. Personal and Ubiquitous Computing, 23, 801–813, 2019. https://doi.org/10.1007/s11042-019-7336-7.
19. Q. Zhu, J. Wang and J. Caverlee, Fairness-aware personalized ranking recommendation via adversarial learning. arXiv preprint, arXiv:2103.07849, 2021. https://doi.org/10.48550/arXiv.2103.07849.
20. E. Yalcin and A. Bilge, Investigating and counteracting popularity bias in group recommendations. Information Processing & Management, 58 (5), 102608, 2021. https://doi.org/10.1016/j.ipm.2021.102608.
21. E. Yalcin and A. Bilge, Treating adverse effects of blockbuster bias on beyond-accuracy quality of personalized recommendations. Engineering Science and Technology, an International Journal, 33, 101083, 2022. https://doi.org/10.1016/j.jestch.2021.101083.
22. M. Chen, A. Beutel, P. Covington, S. Jain, F. Belletti and E. H. Chi, Top-k off-policy correction for a REINFORCE recommender system. Proceedings of the 13th ACM International Conference on Web Search and Data Mining (WSDM '20), 456–464, Houston, TX, USA, 2020.
23. S. Yao and B. Huang, Beyond parity: Fairness objectives for collaborative filtering. Advances in Neural Information Processing Systems, 2017.
24. H. Abdollahpouri, Popularity bias in recommendation: A multi-stakeholder perspective. arXiv preprint, arXiv:2008.08551, 2020. https://doi.org/10.48550/arXiv.2008.08551.
25. Y. Wang, W. Ma, M. Zhang, Y. Liu and S. Ma, A survey on the fairness of recommender systems. ACM Transactions on Information Systems, 40 (3), 1–44, 2022. https://doi.org/10.1145/3547333.
26. Y. Deldjoo, D. Jannach, A. Bellogin, A. Difonzo and D. Zanzonelli, Fairness in recommender systems: Research landscape and future directions. User Modeling and User-Adapted Interaction, 34, 59–108, 2023. https://doi.org/10.1007/s11257-023-09364-z.
27. G. Alves, D. Jannach, R. F. de Souza and M. G. Manzato, User perception of fairness-calibrated recommendations. Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization (UMAP '24), 113–122, Limassol, Cyprus, 2024. https://doi.org/10.1145/3627043.3659558.
28. A. J. B. Chaney, B. M. Stewart and B. E. Engelhardt, How algorithmic confounding in recommendation systems increases homogeneity and decreases utility. Proceedings of the 12th ACM Conference on Recommender Systems (RecSys '18), 224–232, 2018. https://doi.org/10.1145/3240323.3240370.
29. M. Mansoury, H. Abdollahpouri, M. Pechenizkiy, B. Mobasher and R. Burke, Feedback loop and bias amplification in recommender systems. Proceedings of the 29th ACM International Conference on Information & Knowledge Management (CIKM '20), 2145–2148, Virtual Event, Ireland, 2020. https://doi.org/10.1145/3340531.3412152.
30. K. Krauth, Y. Wang and M. I. Jordan, Breaking feedback loops in recommender systems with causal inference. Proceedings of the 39th International Conference on Machine Learning (ICML '22), Baltimore, MD, USA, 2022.
31. S. K. Lam, J. L. Herlocker, J. A. Konstan and J. T. Riedl, Movielens 1M: Collaborative filtering dataset for personalized recommendations. Proceedings of SIGIR, 35–42, 2008.
32. F. M. Harper and J. A. Konstan, The MovieLens datasets: History and context. ACM Transactions on Interactive Intelligent Systems (TiiS), 5 (4), 1–19, 2015. https://doi.org/10.1145/2827872.
33. P. Gopalan, J. M. Hofman and D. M. Blei, Scalable recommendation with hierarchical Poisson factorization. Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence (UAI '15), 326–335, Amsterdam, Netherlands, 2015.
34. N. Srebro and T. Jaakkola, Weighted low-rank approximations. Proceedings of the 20th International Conference on Machine Learning (ICML '03), 720–727, Washington, DC, USA, 2003.
35. D. Liang, R. G. Krishnan, M. D. Hoffman and T. Jebara, Variational autoencoders for collaborative filtering. Proceedings of the 2018 World Wide Web Conference (WWW '18), 689–698, Lyon, France, 2018. https://doi.org/10.1145/3178876.3186150.
36. A. Salah, Q.-T. Truong and H. W. Lauw, Cornac: A comparative framework for multimodal recommender systems. Journal of Machine Learning Research, 21 (95), 1–5, 2020.
37. L. Boratto, G. Fenu, M. Marras and G. Medda, Consumer fairness in recommender systems: Contextualizing definitions and mitigations. Proceedings of the 44th European Conference on Information Retrieval (ECIR '22), 552–566, Stavanger, Norway, 2022.
38. S. M. McNee, J. Riedl and J. A. Konstan, Being accurate is not enough: How accuracy metrics have hurt recommender systems. CHI '06 Extended Abstracts on Human Factors in Computing Systems, 1097–1101, Montréal, Canada, 2006. https://doi.org/10.1145/1125451.1125659.
39. D. C. Silva, M. G. Manzato and F. A. Durão, Exploiting personalized calibration and metrics for fairness recommendation. Expert Systems with Applications, 181, 115112, 2021. https://doi.org/10.1016/j.eswa.2021.115112.
40. R. Sanders, The Pareto principle: Its use and abuse. Journal of Services Marketing, 1 (2), 37–40, 1987.
41. P. J. Chia, J. Tagliabue, F. Bianchi, C. He and B. Ko, Beyond NDCG: Behavioral testing of recommender systems with RecList. Companion Proceedings of the Web Conference 2022 (WWW '22), 99–104, Virtual Event, 2022.
42. J. Bobadilla, F. Ortega, A. Hernando and J. Bernal, A collaborative filtering approach to mitigate the new user cold start problem. Knowledge-Based Systems, 26, 225–238, 2012. https://doi.org/10.1016/j.knosys.2011.07.021.
43. S. Wang, M. Gong, H. Li and J. Yang, Multi-objective optimization for long tail recommendation. Knowledge-Based Systems, 104, 145–155, 2016. https://doi.org/10.1016/j.knosys.2016.04.018.
44. M. Elahi, B. Ferwerda, M. Tkalčič, M. Schedl and F. Ricci, Investigating the impact of recommender systems on user-based and item-based popularity bias. Information Processing & Management, 58 (5), 102655, 2021. https://doi.org/10.1016/j.ipm.2021.102655.