Privacy-Preserving Recommendation with Debiased Obfuscaiton
Abstract
As people enjoy the personalized services recommended by Recommender Systems (RSs), the privacy disclosure risk increases with frequent interactions. Malicious adversary often collects public information online to infer private information for illicit profit. As privacy concerns grew, researchers introduced data obfuscation into recommender systems. However, there still exists several limitations in current work. First, although the existing methods effectively reduce the risk of privacy disclosure, they can be detrimental to the quality of the recommendation service. Second, a range of practical issues under the application of recommendation systems are not considered, e.g., long-tail, density, etc. To address those challenges, we propose a novel framework named Want User Defending Inference (WUDI), a high-performance privacy-preserving debiased framework based on data obfuscation. Unlike the original strategies, i.e., adding or removing user ratings, we introduced some novel strategies to generate an obfuscated matrix. Firstly, we define a new method called Cluster Recommend for alleviating the long-tail skewness and data sparsity in RSs. Then we investigate the gender bias in obfuscation and apply a bias mitigating strategy to RSs. Experiments on public datasets demonstrate that WUDI can outperform the state-of-the-art baselines in obfuscation.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Obfuscation mitigates data privacy disclosure risk by distorting the data by adding, removing, or altering existing data in a way that blocks the adversary from conducting an inference attack and also ensures recommendations accuracy as it contains no such noisy data.
“In the obfuscation approach, the data privacy disclosure risk is mitigated by distorting the data by adding, removing, or altering existing data in a way so that the whole process blocks the adversary from conducting an inference attack and also ensures recommendations accuracy as it contains no such noisy data [118].”
From A Deep Dive into Fairness, Bias, Threats, and Privacy in Recommender Systems: Insights and Future Research (2024) · cites this paper as [118]
Obfuscation can be achieved by solving a multi-objective optimization problem where the objectives are improving RS accuracy and protecting users' privacy, or applying probabilistic privacy mapping between private and to-be-released public data, or removing the most indicative items of the sensitive data with the addition of some fictitious data to confuse the adversary.
“The obfuscation can be achieved by solving the multi-objective optimization problem where the objectives are improving RS accuracy & protecting users' privacy [119] or applying probabilistic privacy mapping between private & to be released public data [100] or removing the most indicative items of the sensitive data with the addition of some fictitious data to confuse the adversary [10, 11”
From A Deep Dive into Fairness, Bias, Threats, and Privacy in Recommender Systems: Insights and Future Research (2024) · cites this paper as [118]
Lin et al. [118] addressed the same issue while accounting for popularity bias by incorporating long-tail (unpopular) items.
“Subsequently, Lin et al. [118] addressed the same issue while accounting for popularity bias by incorporating long-tail (unpopular) items.”
From A Deep Dive into Fairness, Bias, Threats, and Privacy in Recommender Systems: Insights and Future Research (2024) · cites this paper as [118]
Cites
In-Edge AI: Intelligentizing Mobile Edge Computing, Caching and Communication by Federated Learning
Wang, X., Han, Y., Wang, C., Zhao, Q., Chen, X., & Chen, M. (2019) · IEEE Network · doi:10.1109/mnet.2019.1800286 · Cited by 1,030 · cited 1× · see citing works
“It can generate more robust models without sharing data across devices, resulting in privacy-preserving solutions with higher security, making it especially appropriate for privacy-sensitive environments [13], [14].”
Background"You Might Also Like:" Privacy Risks of Collaborative Filtering
Calandrino, J. A., Kilzer, A., Narayanan, A., Felten, E. W., & Shmatikov, V. (2011) · 2011 IEEE Symposium on Security and Privacy · doi:10.1109/sp.2011.40 · Cited by 311 · see citing works
Slokom, M., Hanjalic, A., & Larson, M. (2021) · Information Processing & Management · doi:10.1016/j.ipm.2021.102722 · Cited by 31 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Roy, F., Ding, X., Choo, K. K. R., & Zhou, P. (2024) · arXiv · doi:10.48550/arxiv.2409.12651
- Authors
- Lin, C., Liu, B., Zhang, X., Wang, Z., Hu, C., & Luo, L.
- Affiliations
- Ningbo University (CN)
- Venue
- 2022 IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom)
- Publisher
- IEEE
- Published
- 2022-12
- Pages
- 590-597
- OpenAlex
- W4328028476
- Cited by
- 4
- References
- 3
- Citations by year
- 2025: 2 · 2024: 1 · 2023: 1
- FWCI
- 0.55
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom) (publishedVersion)
- Funders
- Natural Science Foundation of Zhejiang Province
- Science and Technology Innovation 2025 Major Project of Ningbo
- Natural Science Foundation of Ningbo
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
- Physical Sciences › Computer Science › Information Systems › Recommender Systems and Techniques
- Social Sciences › Sociology and Political Science › Privacy, Security, and Data Protection
- Keywords
- privacy-preserving recommendation, data obfuscation, recommender systems, data sparsity, gender bias
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
