Differentially Private SQL with Bounded User Contribution
This work addresses a practical gap in differential privacy systems for real-world databases where user contributions are unbounded, though it is incremental in extending existing methods to handle multiple records per individual.
The paper tackled the problem of performing differentially private aggregations on databases where individuals can be associated with arbitrarily many rows, a limitation in existing systems, and implemented a scalable method in an SQL engine, validating it with industry benchmarks and open-sourcing the components.
Differential privacy (DP) provides formal guarantees that the output of a database query does not reveal too much information about any individual present in the database. While many differentially private algorithms have been proposed in the scientific literature, there are only a few end-to-end implementations of differentially private query engines. Crucially, existing systems assume that each individual is associated with at most one database record, which is unrealistic in practice. We propose a generic and scalable method to perform differentially private aggregations on databases, even when individuals can each be associated with arbitrarily many rows. We express this method as an operator in relational algebra, and implement it in an SQL engine. To validate this system, we test the utility of typical queries on industry benchmarks, and verify its correctness with a stochastic test framework we developed. We highlight the promises and pitfalls learned when deploying such a system in practice, and we publish its core components as open-source software.