Abstract:Recommender systems rely on large datasets of historical data and entail serious privacy risks. A server offering Recommendation as a Service to a client might leak more information than necessary regarding its recommendation model and dataset. At the same time, the disclosure of the client's preferences to the server is also a matter of concern. Devising privacy-preserving protocols using general cryptographic primitives (e.g., secure multi-party computation or homomorphic encryption), is a typical approach to overcome privacy concerns, but in conjunction with state-of-the-art recommender systems often yields far-from-practical solutions. In this paper, we tackle this problem from the direction of constructing crypto-friendly machine learning algorithms. In particular, we propose CryptoRec, a secure two-party computation protocol for Recommendation as a Service, which encompasses a novel recommender system. This model possesses two interesting properties: (1) It models user-item interactions in an item-only latent feature space in which personalized user representations are automatically captured by an aggregation of pre-learned item features. This means that a server with a pre-trained model can provide recommendations for a client whose data is not in its training set. Nevertheless, re-training the model with the client's data still improves accuracy. (2) It only uses addition and multiplication operations, making the model straightforwardly compatible with homomorphic encryption schemes. We demonstrate the efficiency and accuracy of CryptoRec on three real-world datasets. CryptoRec allows a server with thousands of items to privately answer a prediction query within a few seconds on a single PC, while its prediction accuracy is still competitive with state-of-the-art recommender systems computing over clear data.

A Secure And Efficient Framework For Privacy Preserving Social Recommendation

Privacy-preserving recommendation system based on social relationships

A Novel Social Recommendation Method Fusing User’s Social Status and Homophily Based on Matrix Factorization Techniques

Preserving Data Privacy in Social Recommendation

An Efficient Clustering-Based Privacy-Preserving Recommender System.

Hiding Your Awful Online Choices Made More Efficient and Secure: A New Privacy-Aware Recommender System

Secure Social Recommendation Based on Secret Sharing

A Privacy-Preserving Framework for Trust-Oriented Point-of-Interest Recommendation

An Algorithm for Efficient Privacy-Preserving Item-Based Collaborative Filtering

Efficient Privacy-Preserving Content Recommendation for Online Social Communities

An Effective Federated Recommendation Framework with Differential Privacy

Privacy-Preserving Federated Cross-Domain Social Recommendation.

A Robust Reputation-Based Location-Privacy Recommender System Using Opportunistic Networks.

Advanced Privacy-Preserving Federated Relationship Recommendation

Holistic Collaborative Privacy Framework for Users' Privacy in Social Recommender Service

Exploiting Data Sparsity in Secure Cross-Platform Social Recommendation

Privacy-Preserving Friend Recommendation in an Integrated Social Environment

CryptoRec: Privacy-preserving Recommendation as a Service

Pps-Poi-Rec: A Privacy Preserving Social Point-Of-Interest Recommender System

A Scalable Algorithm for Privacy-Preserving Item-based Top-N Recommendation

DPLCF: Differentially Private Local Collaborative Filtering