Two Differentially Private Rating Collection Mechanisms for Recommender Systems

TL;DR

This paper introduces two differentially private mechanisms for collecting user ratings in recommender systems, ensuring privacy while maintaining data utility.

Contribution

It proposes two novel mechanisms, a modified Laplace and a randomized response, that are both differentially private and utility-preserving.

Findings

Both mechanisms are proven to be differentially private.

They effectively preserve data utility in rating collection.

The mechanisms outperform traditional methods in privacy-utility trade-offs.

Abstract

We design two mechanisms for the recommender system to collect user ratings. One is modified Laplace mechanism, and the other is randomized response mechanism. We prove that they are both differentially private and preserve the data utility.