Bayesian Local Differential Privacy for Implicit Feedback Recommendation

Hao Tang; Yong Wang; Bo Li; Jiangzhou Deng; Zhiqiang Zhang

Abstract

Recommender systems have become essential tools for personalized information delivery, yet their reliance on user interaction data raises significant privacy concerns. Local differential privacy (LDP) provides strong privacy guarantees by perturbing user data before it leaves the client, but the injected noise often severely degrades recommendation accuracy. To address this challenge, we propose a novel LDP-based recommendation algorithm using Bayesian estimation. The method first perturbs users’ implicit feedback locally using a randomized response mechanism, then reconstructs the true user–item interaction probabilities on the server through Bayesian inference. This two-step approach effectively mitigates noise while preserving privacy, enabling high-quality model training even under strict privacy constraints. Extensive experiments on three public datasets demonstrate that our method achieves superior recommendation performance compared with state-of-the-art algorithms, striking a favorable balance between privacy protection and utility. This study provides a practical and scalable solution for privacy-preserving recommendations, particularly in scenarios involving untrusted servers and sparse.

References

1.
Ge, Y.; Liu, S.; Fu, Z.; et al. A survey on trustworthy recommender systems. ACM Trans. Recomm. Syst. 2024, 3, 1–68.
2.
Xu, L.; Zhang, J.; Li, B.; et al. Tapping the potential of large language models as recommender systems: A comprehensive framework and empirical analysis. ACM Trans. Knowl. Discov. Data 2024, 19, 105.
3.
Alzahrani, A.; Asghar, M.Z. Maintaining user security in consumer electronics-based online recommender systems using federated learning. IEEE Trans. Consum. Electron. 2024, 70, 2657–2665.
4.
Deng, J.; Wu, Q.; Wang, S.; et al. A novel joint neural collaborative filtering incorporating rating reliability. Inf. Sci. 2024, 665, 120406.
5.
Rendle, S.; Freudenthaler, C.; Gantner, Z.; et al. BPR: Bayesian personalized ranking from implicit feedback. arXiv 2012, arXiv:1205.2618.
6.
Lin, S.; Zhou, S.; Chen, J.; et al. ReCRec: Reasoning the causes of implicit feedback for debiased recommendation. ACM Trans. Inf. Syst. 2024, 42, 158.
7.
Zhu, H.; Xiong, F.; Chen, H.; et al. Incorporating a triple graph neural network with multiple implicit feedback for social recommendation. ACM Trans. Web 2024, 18, 23.
8.
Liu, H.; Wang, Y.; Zhang, Z.; et al. Matrix factorization recommender based on adaptive Gaussian differential privacy for implicit feedback. Inf. Process. Manag. 2024, 61, 103720.
9.
Luo, C.; Wang, Y.; Zhang, Y.; et al. Distributed differentially private matrix factorization for implicit data via secure aggregation. IEEE Trans. Comput. 2024, 74, 705–716.
10.
Zhang, S.; Yin, H. Comprehensive privacy analysis on federated recommender system against attribute inference attacks. IEEE Trans. Knowl. Data Eng. 2022, 36, 987–999.
11.
Zhu, Z.; Wu, C.; Fan, R.; et al. Membership inference attacks against sequential recommender systems. In Proceedings of the ACM Web Conference 2023; Association for Computing Machinery: New York, NY, USA, 2023.
12.
Dwork, C.; McSherry, F.; Nissim, K.; et al. Calibrating noise to sensitivity in private data analysis. In Third Theory of Cryptography Conference (TCC 2006); Springer: Berlin/Heidelberg, Germany, 2006; pp. 265–284.
13.
Dwork, C.; Roth, A. The algorithmic foundations of differential privacy. Found. Trends Theor. Comput. Sci. 2014, 9, 211–407.
14.
McSherry, F.; Mironov, I. Differentially private recommender systems: building privacy into the Netflix prize contenders. In Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Paris, France, 28 June–1 July 2009; pp. 627–636.
15.
Xu, Z.; Chu, C.; Song, S. An effective federated recommendation framework with differential privacy. Electronics 2024, 13, 1589.
16.
Deng, Y.; Zhou, W.; Haq, A.U.; et al. Differentially private recommender framework with dual semi-autoencoder. Expert Syst. Appl. 2025, 260, 125447.
17.
Di Fazio, A. Enhancing privacy in recommender systems through differential privacy techniques. In Proceedings of the 18th ACM Conference on Recommender Systems, Bari, Italy, 14–18 October 2024; pp. 1348–1352.
18.
Demelius, L.; Kern, R.; Tru… gler, A. Recent advances of differential privacy in centralized deep learning: A systematic survey. Acm Comput. Surv. 2025, 57, 158.
19.
Guo, T.; Luo, J.; Dong, K.; et al. Locally differentially private item-based collaborative filtering. Inf. Sci. 2019, 502, 229–246.
20.
Gao, C.; Huang, C.; Lin, D.; et al. DPLCF: differentially private local collaborative filtering. In Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, Virtual, 25–30 July 2020; pp. 961–970.
21.
Hu, W.; Fang, H. Towards differential privacy in sequential recommendation: a noisy graph neural network approach. Acm Trans. Knowl. Discov. Data 2024, 18, 125.
22.
Ferrara, A.; Fazio, A.D.; Mancino, A.C.M.; et al. Enhancing utility in differentially private recommendation data release via exponential mechanism. In European Conference on Information Retrieval; Springer Nature: Cham, Switzerland, 2025; pp. 34–51.
23.
Neera, J.; Chen, X.; Aslam, N.; et al. Private and utility enhanced recommendations with local differential privacy and Gaussian mixture model. IEEE Trans. Knowl. Data Eng. 2021, 35, 4151–4163.
24.
Sarkar, S.; Shinde, S.; Shedge, R. Elevating privacy in recommendation systems with hybrid noise in local differential privacy. In World Conference on Artificial Intelligence: Advances and Applications; Springer Nature: Singapore, 2024; pp. 241–256.
25.
Mu… llner, P.; Lex, E.; Schedl, M.; et al. The impact of differential privacy on recommendation accuracy and popularity bias. In European Conference on Information Retrieval; Springer Nature: Cham, Switzerland, 2024; pp. 466–482.
26.
Rezaimehr, F.; Dadkhah, C. A survey of attack detection approaches in collaborative filtering recommender systems. Artif. Intell. Rev. 2020, 54, 2011–2066.
27.
Zhang, M.; Ren, Z.; Wang, Z.; et al. Membership inference attacks against recommender systems. In Proceedings of the 2021 ACM SIGSAC Conference on Computer and Communications Security, Virtual, 15–19 November 2021; pp. 864–879.
28.
Himeur, Y.; Sohail, S.S.; Bensaali, F.; et al. Latest trends of security and privacy in recommender systems: A comprehensive review and future perspectives. Comput. Secur. 2022, 118, 102746.
29.
Pan, Y.; Wu, Z.S.; Xu, H.; et al. Differentially private Bayesian persuasion. In Proceedings of the ACM on Web Conference 2025; Association for Computing Machinery: New York, NY, USA, 2025; pp. 1425–1440.
30.
Fienberg, S.E. A brief history of statistical models for network analysis and open challenges. J. Comput. Graph. Stat. 2012, 21, 825–839.
31.
Chatterjee, S.; Diaconis, P.; Sly, A. Random graphs with a given degree sequence. Ann. Appl. Probab. 2010, 21, 1400–1435.
32.
Du, M.; Peng, J.; Hu, Y.;et al. Logistic regression matrix factorization recommendation algorithm for differential privacy. J. Beijing Univ. Posts Telecommun. 2023, 46, 115–120. (In Chinese)
33.
He, X.; Deng, K.; Wang, X.; et al. LightGCN: Simplifying and powering graph convolution network for recommendation. In Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, Virtual, 25–30 July 2020; pp. 639–648.
34.
Ran, X.; Wang, Y.; Zhang, L.; et al. A differentially private nonnegative matrix factorization for recommender system. Inf. Sci. 2022, 592, 21–35.

Scilight Press

Author Information

Abstract

Keywords

References

About Scilight

Journals

Publishing Policies

Contact Us