Distributed Differentially Private Ranking Aggregation

TL;DR

This paper introduces a distributed differential privacy mechanism for ranking aggregation that enhances privacy protection in untrusted curator scenarios while maintaining high accuracy.

Contribution

It proposes a novel distributed differential privacy framework with a shuffle model for ranking aggregation, addressing untrusted curators and improving privacy and accuracy.

Findings

The mechanism provides strong privacy guarantees under the shuffle model.

Experimental results show competitive accuracy compared to existing methods.

Theoretical analysis confirms the privacy amplification effect.

Abstract

Ranking aggregation is commonly adopted in cooperative decision-making to assist in combining multiple rankings into a single representative. To protect the actual ranking of each individual, some privacy-preserving strategies, such as differential privacy, are often used. This, however, does not consider the scenario where the curator, who collects all rankings from individuals, is untrustworthy. This paper proposed a mechanism to solve the above situation using the distribute differential privacy framework. The proposed mechanism collects locally differential private rankings from individuals, then randomly permutes pairwise rankings using a shuffle model to further amplify the privacy protection. The final representative is produced by hierarchical rank aggregation. The mechanism was theoretically analysed and experimentally compared against existing methods, and demonstrated competitive results in both the output accuracy and privacy protection.