A Hybrid Matrix Factorization Framework for Balancing Personalization, Diversity, and Coverage in Ad Recommendation Systems
DOI:
https://doi.org/10.64229/cas25k91Keywords:
Hybrid Recommender Systems, Personalization, Diversity, Coverage, Advertisement Recommendation, Matrix Factorization, Fairness, SVD, NMF, User-Centric DesignAbstract
This paper presents a comprehensive framework for addressing the inherent trade-offs among personalization, diversity, and coverage in hybrid advertisement recommendation systems. In response to the growing complexity of online advertising, the proposed system integrates collaborative filtering via Singular Value Decomposition (SVD) with content-based filtering using Non-negative Matrix Factorization (NMF), enhanced by a re-ranking mechanism based on coverage profiles. This hybrid design aims to deliver relevant, diverse, and widely distributed ads, thereby improving user engagement and fairness among content providers. To quantitatively evaluate these objectives, three novel metrics—Ad User Profile (AUP), Ad Candidate Set (ACS), and Ad Matching Degree (AMD)—are introduced. Through empirical analysis, including visualization and comparative experiments, the study demonstrates that the proposed system outperforms traditional models in achieving a more balanced recommendation outcome. Additionally, real-world case studies and optimization formulations are explored to support scalable deployment. This work contributes to the evolving landscape of recommender systems by proposing a scalable, user-centric approach that harmonizes personalization with fairness and discovery in digital advertising.
References
[1]G. Adomavicius and A. Tuzhilin, “Context-aware recommender systems,” in Recommender Systems Handbook, Springer, 2015, pp. 191–226.
[2]Y. Koren, R. Bell, and C. Volinsky, “Matrix factorization techniques for recommender systems,” Computer, vol. 42, no. 8, pp. 30–37, 2009.
[3]M. Zhang and N. Hurley, “Avoiding monotony: Improving the diversity of recommendation lists,” in Proc. ACM RecSys, 2008, pp. 123–130.
[4]Y. Zhao et al., “Fairness and diversity in recommender systems: A survey,” arXiv preprint, arXiv:2307.04644, 2023.
[5]T. Zhou et al., “Solving the apparent diversity-accuracy dilemma of recommender systems,” PNAS, vol. 107, no. 10, pp. 4511–4515, 2010.
[6]D. Jannach and L. Lerche, “From personalized to diversity-aware recommendation: A case study in e-commerce,” in Proc. ACM RecSys, 2017, pp. 32–40.
[7]E. Çano and M. Morisio, “Hybrid recommender systems: A systematic literature review,” arXiv preprint, arXiv:1901.03888, 2019.
[8]Y. Hu, Y. Koren, and C. Volinsky, “Collaborative filtering for implicit feedback datasets,” in Proc. IEEE ICDM, 2008, pp. 263–272.
[9]G. Lops, M. de Gemmis, and G. Semeraro, “Content-based recommender systems: State of the art and trends,” in Recommender Systems Handbook, Springer, 2011, pp. 73–105.
[10]D. Lee and H. Seung, “Algorithms for non-negative matrix factorization,” in Advances in Neural Information Processing Systems, 2001.
[11]P. Covington, J. Adams, and E. Sargin, “Deep neural networks for YouTube recommendations,” in Proc. ACM RecSys, 2016, pp. 191–198.
[12]D. Chakrabarti, R. Kumar, and F. Radlinski, “Adversarial click prediction,” in Proc. WSDM, 2008, pp. 7–16.
[13]S. Wu et al., “Multistakeholder recommendation: Applications and challenges,” ACM Trans. Intell. Syst. Technol., vol. 10, no. 4, pp. 1–32, 2019.
[14]X. He et al., “Practical lessons from predicting clicks on ads at Facebook,” in Proc. ADKDD, 2014.
[15]E. Pariser, The Filter Bubble: What the Internet Is Hiding from You. Penguin UK, 2011.
[16]Y. Koren, “Factorization meets the neighborhood: a multifaceted collaborative filtering model,” in Proc. ACM KDD, 2008, pp. 426–434.
[17]H. Cheng et al., “Wide & deep learning for recommender systems,” in Proc. ACM RecSys, 2016, pp. 7–10.
[18]S. Rendle and L. Schmidt-Thieme, “Online-updating regularized kernel matrix factorization models for large-scale recommender systems,” in Proc. ACM RecSys, 2008, pp. 251–258.
[19]J. Carbonell and J. Goldstein, “The use of MMR, diversity-based reranking for reordering documents and producing summaries,” in Proc. ACM SIGIR, 1998, pp. 335–336.
[20]L. Wei and E. Gabrilovich, “Using large-scale graph-based recommendation in ad targeting,” in Proc. CIKM, 2014, pp. 1827–1830.
[21]A. Kulesza and B. Taskar, “Determinantal point processes for machine learning,” Foundations and Trends in Machine Learning, vol. 5, no. 2–3, pp. 123–286, 2012.
[22]J. Tintarev and P. L. Brusilovsky, “Providing control and transparency in social recommendation: A position paper,” in Proc. RecSys Workshop on Human Decision Making, 2011.
[23]R. Burke, “Multisided fairness for recommendation,” in Proc. FATREC Workshop on Responsible Recommendation, 2017. review,” arXiv preprint, arXiv:1901.03888, 2019.
[24]D. Silver et al., “Reinforcement learning with function approximation for recommendations,” in Proc. ACM RecSys, 2013, pp. 33–40.
[25]S. Mehrotra et al., “Towards a fair marketplace: Counterfactual evaluation of the trade-off between relevance, fairness & satisfaction in recommendation systems,” in Proc. FAT, 2018.
[26]J. Beel, S. Langer, M. Genzmehr, and A. Nürnberger, “Personalized ranking for scientific literature recommendation: Comparing different ranking factors,” in Proc. i-KNOW, 2011.
[27]D. Holtz et al., “The engagement-diversity connection: Evidence from a field experiment on Spotify,” in Proc. ACM WWW, 2020, pp. 1–12.
[28]S. Wang et al., “Exposure fairness in recommendation: From exposure disparity to group imbalance,” in Proc. ACM RecSys, 2022, pp. 21–29.
[29]J. Zhang et al., “Fairness-aware recommender systems: A survey,” ACM Comput. Surv., vol. 54, no. 5, pp. 1–38, 2021.
[30]C. Vargas and P. Castells, “Rank and relevance in novelty and diversity metrics for recommender systems,” in Proc. ACM RecSys, 2011, pp. 109–116.
[31]A. Ekstrand and M. Pera, “Balancing accuracy and diversity in recommender systems with Pareto-efficient rankings,” in Proc. ACM FAT, 2019, pp. 90–98.
[32]P. Patro et al., “FairRec: Two-sided fairness for personalized recommendations in two-sided platforms,” in Proc. ACM RecSys, 2020, pp. 598–607.
[33]J. Kaminskas and F. Ricci, “Personalized music recommendation: A multidimensional approach,” in Proc. RecSys, 2011, pp. 275–278.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Yaqeen Mousa Abid, Ali Fahem Neamah (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
