Multi-Analyst Differential Privacy with Fine-Grained Provenance for Databases
Xi Guang He · ACM SIGMOD Record · 2025
Motivation. Differential Privacy (DP) [8] has emerged as a promising standard for safeguarding the privacy of data contributors when their information is used in data-driven applications and research. Answering queries using DP has a bounded information disclosure of the data contributors to the data analyst. This information disclosure is quantified by a privacy parameter in DP, also known as the privacy budget. In practice, internal data analysts or applications with a higher privilege level for critical tasks like security alerts might be granted a bigger privacy budget and, hence, more extensive access to sensitive data, than an external application such as third-party advertisements.