This work presents and evaluates a privacy-preserving recommendation framework that combines federated learning, differential privacy, cohort-level modeling, and privacy-aware intelligent agents, offering a scalable approach for next-generation AI-driven retail platforms.
Abstract
Personalized recommendation systems are central to modern e-commerce and retail platforms, but they typically rely on centralized storage of detailed user interaction data, creating significant privacy and regulatory challenges. With increasing requirements from regulations such as GDPR, CCPA, and CPRA, organizations must develop recommendation systems that preserve user privacy without substantially degrading recommendation quality. This work presents and evaluates a privacy-preserving recommendation framework that combines federated learning, differential privacy, cohort-level modeling, and privacy-aware intelligent agents. The framework keeps raw user data decentralized while introducing mathematically bounded noise to model updates. Experiments were conducted on synthetic retail datasets that emulate customer clickstream and purchase behavior. Recommendation quality was evaluated using Click-Through Rate (CTR), Precision@K, Recall@K, and Normalized Discounted Cumulative Gain (NDCG@K) across multiple differential privacy budgets. We evaluate matrix factorization, neural collaborative filtering, and GRU4Rec under varying privacy constraints and analyze the trade-off between privacy and utility. An interactive Streamlit dashboard was developed to visualize recommendation performance, ranking stability, privacy-utility trade-offs, and fairness metrics. Results show that the proposed framework maintains competitive recommendation quality at moderate privacy budgets (approximately $\epsilon \approx 5$), demonstrating that strong privacy guarantees can be achieved with limited impact on recommendation effectiveness. This work provides a practical framework for deploying privacy-preserving recommendation systems that balance personalization, regulatory compliance, and business objectives, offering a scalable approach for next-generation AI-driven retail platforms.
Federated Recommender Systems (FRSs) enable on-device personalization while preserving user privacy, yet the gradients shared during model training can still leak sensitive information. Local differential privacy (LDP) offers a formal defense, but classic mechanisms—randomized response and Laplace perturbation—inject substantial noise into the high-dimensional gradient vectors of latent-factor models, severely degrading recommendation accuracy. In this paper, we systematically evaluate these traditional LDP strategies in federated matrix factorization under both explicit (MovieLens 100K and 1M) and implicit (Steam, LastFM) feedback benchmarks. We then propose a novel directional noise mechanism based on the von Mises–Fisher distribution, which preserves gradient magnitudes while randomizing directions. Through extensive experiments across a wide range of privacy budgets, we show that our vMF mechanism consistently outperforms Laplace and binary randomized-response, recovering up to more than 100 % of the non-private utility at ε = 1 and delivering higher ranking accuracy under moderate privacy levels (ε ≤ 2). We further demonstrate that directional perturbation reduces variance and remains robust under partial client participation. Our findings highlight the importance of data geometry in privacy–utility trade-offs and provide practical guidance for deploying LDP in high-dimensional federated recommendation settings.
Nawfal Abbassi Saber, Loubna Mekouar, Youssef Iraqi· European Symposium on Securi...· 0 citations
Recommendation systems play an important role in helping users to discover relevant items out of large collections of content. However, the widespread use of information about people's personal preferences raises concerns about privacy. In this research paper, we examine the combinatorial testing strategies to enhance privacy protection in recommendation engines without the detriment of prediction performance. The approach proposed includes a systematic evaluation of combinations of recommendation algorithms and privacy mechanisms by using the MovieLens 20M dataset. The data set has 20,000,263 user ratings and 465,564 tag applications on 27,278 movies created by 138,493 users between January 1995 and March 2015, which currently serves as a popular benchmark for recommender system research. The proposed combinatorial privacy testing model is compared with collaborative filtering, matrix factorization and deep neural recommendation methods. Experimental results show that the performance of the proposed model reaches better results with accuracy of 94.18%, precision of 93.40%, recall of 92.96% and F1-score of 93.18%. The results demonstrate that combinatorial privacy test can be effectively used to increase the reliability of recommendations or protect the information of users.
Prerana Nilesh Khairnar, Gujjala Srinath, A. B. Pawar et al.· Journal of Intelligent Decis...· 0 citations
The results demonstrate that federated learning is a scalable and effective method that can achieve privacy compliance in e-commerce analytics within data-restricted environments, and it lays a solid foundation for secure distributed business intelligence.
Jing Hao· International Conference on...· 0 citations
Differential privacy (DP) mechanisms have been widely adopted in federated learning (FL) to enhance model security. However, existing approaches predominantly employ uniform privacy budgets, neglecting personalized requirements arising from heterogeneous user privacy preferences. Such uniform privacy configurations typically necessitate compliance with the most stringent budget, which not only leads to the wasteful underutilization of privacy budgets for certain clients but also compromises overall model utility. To address this limitation, we propose FedSPA, a Subspace Projection Aggregation personalized differential private Federated learning framework. The proposed method conducts singular value decomposition operations on noise-perturbed local models to extract singular value vectors as compact representations of both model structure and privacy noise. The server then clusters clients and identifies a consensus subspace for projecting models with varying noise levels, ultimately aggregating the global model through a residual-aware mechanism. This method not only effectively guides the aggregation of client personalized differential privacy but also reduces communication overhead. Extensive experiments demonstrate the model's effectiveness. Additionally, we provide theoretical proof of the privacy and convergence of FedSPA. Experimental results also showcase its superior performance over personalized DP-FL baselines.
Tianchi Liao, Xiaojun Deng, Lele Fu et al.· Proceedings of the 32nd ACM...· 0 citations
Recommender systems personalize digital experiences across e-commerce, streaming, and social media, yet they face linked challenges in fairness, bias, threats, and privacy. Bias can lead to unequal treatment, so fairness seeks equitable recommendations for all stakeholders. At the same time, adversarial attacks can erode reliability, and extensive use of personal data creates privacy risks that demand robust safeguards. This study surveys these challenges, shows how algorithmic choices can unintentionally reinforce disparities, and reviews strategies for fair recommendation and privacy-preserving learning. We frame these aspects with a unified Privacy–Accuracy–Fairness plus Threat framework that spans data retrieval, ranking and serving, and we advocate standardized evaluation on fixed candidate item pools to isolate tradeoffs between quality and responsibility metrics. By integrating these dimensions, we synthesize current limitations and outline future directions to improve threat resilience, fairness and privacy. We also situate the discussion in ethical and regulatory contexts and translate those principles into practical controls that foster trustworthy systems for diverse populations.
PADP is presented, a sensitivity-aware perturbation framework inspired by differential privacy principles, which provides a plug-and-play, middleware framework that can be easily integrated into enterprise RAG pipelines without requiring costly computations for LLM fine-tuning and reconstruction of vector indices.
Seçkin Mandaci, Yılmaz Vural, Ö. Turna· Black Sea Journal of Enginee...· 0 citations