Privacy-Preserving Personalized Recommender Systems
Bibliographic record
Abstract
Problem Definition: Personalized product recommendations are crucial for online platforms but pose privacy risks. To address these concerns, we propose recommendation policies that adhere to differential privacy constraints. Methodology and Results: We develop a theoretical model where the recommendation policy selects products based on consumers' preference rankings, learned from personal data. Unlike conventional recommendation policies that primarily focus on prospering from meeting consumer satisfaction, our approach applies differential privacy to mitigate the risk of exposing personal information to man-in-the-middle attackers during the transmission of recommendations over communication networks, such as the Internet. As a result, this policy accounts for the trade-off between personalization and privacy. Our analysis shows that the optimal policy is a coarse-grained threshold policy, where products are randomly recommended with either high or low probability based on whether their preference rankings are above or below a certain threshold. We further explore the comparative statics of this threshold in an asymptotic regime with a large number of products, as is typical for online platforms. Moreover, we examine the economic implications of privacy protection. When product prices are exogenous, privacy protection reduces consumer surplus due to lower match values between consumers and recommended products. However, when retailers set prices endogenously, the impact on consumer surplus is non-monotonic, reflecting a trade-off between recommendation accuracy and price inflation. Managerial and Regulatory Implications: Our findings offer valuable insights for practitioners developing privacy-preserving personalized recommendation policies and provide regulators with a deeper understanding of the economic consequences of privacy protection in recommender systems.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.024 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.001 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.006 |
| Open science | 0.004 | 0.003 |
| Research integrity | 0.003 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".