Treffer: R2T: Instance-optimal Truncation for Differentially Private Query Evaluation with Foreign Keys.

Title:
R2T: Instance-optimal Truncation for Differentially Private Query Evaluation with Foreign Keys.
Authors:
Dong, Wei1 (AUTHOR) wei_dong@ntu.edu.sg, Fang, Juanru2 (AUTHOR) jfangad@cse.ust.hk, Yi, Ke2 (AUTHOR) yike@cse.ust.hk, Tao, Yuchao3 (AUTHOR) yctao@cs.duke.edu, Machanavajjhala, Ashwin3 (AUTHOR) ashwin@cs.duke.edu
Source:
Communications of the ACM. Mar2025, Vol. 68 Issue 3, p93-101. 9p.
Subject Terms:
*RELATIONAL databases, *ALGORITHMS, QUERYING (Computer science), DATA privacy, SUBGRAPHS
Database:
Business Source Premier

Weitere Informationen

This research article presents Race-to-the-Top (R2T), a differential privacy mechanism for relational databases. This mechanism builds upon the work done by Kotsogiannis et al. on the utilization of the truncation mechanism and foreign keys. Here, the authors introduced R2T with instance-optimal truncation taking into consideration self-joins, global sensitivity and worse-case optimality. Then, a demonstration using graph pattern-counting queries and real-world networks datasets is presented.