Personalized Privacy-Preserving Semi-Centralized Recommendation System in a Trust-Based Agent Network
Bibliographic record
Abstract
In the current big data era, recommendation systems play an important role in our daily life to help us make faster and better decisions from massive numbers of choices. Personalized recommendation has gained its popularity as it provides recommendations according to the user profile, preferences and/or interests. Many existing systems make recommendation by centralizing data. However, the exposure of sensitive data raises a privacy concern as research has shown that it is possible to de-identify anonymous users. Examples include inferring sensitive information (e.g., political views, sexual orientations) from non-sensitive data (e.g., movie ratings). In this paper, we present a personalized privacy-preserving recommendation system called Trust-based Agent Network (TAN). It tackles the privacy issue by semi-decentralizing data and treating each node in the network as an agent. As such, data are distributed to each agent within each trusted network, and the recommendation service provider collects only obfuscated data from agents by adopting the differential-privacy mechanism. Consequently, data in our TAN are either protected inside local trusted networks or obfuscated outside of trusted networks. Final recommendation can then be made by aggregating the local suggestions from the trusted network and obfuscated global suggestions from the service provider. Personalized recommendations can be made by putting more emphasize on local suggestions. Evaluation results show that our TAN leads to high accuracy and highly personalized recommendations while protecting privacy.
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 | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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".