Seminar on Private Data Systems

The seminar explores Private Database Systems with an emphasis on Secure Collaborative Analysis (SCA) using Secure Multi-Party Computation (MPC). The seminar bridges modern database systems and applied cryptography, examining how analytical queries can be executed across mutually distrusting parties without revealing raw data. The seminar progresses from fundamentals to research-level topics. Students will first understand the required cryptographic primitives (e.g., MPC protocols) and how standard relational query operators (e.g., selection, projection, joins, aggregation) are translated into MPC primitives. From there, the seminar advances toward query optimization techniques tailored for MPC execution, including cost modeling and query planning. The format is research-driven: students will read, present, and critically review key papers in the field. The seminar emphasizes deep technical understanding, analytical thinking, and the ability to evaluate research contributions. For the first part of the seminar, the instructor will present the cryptographic primitives and lead the discussion. In the second part of the seminar, students will take the lead in presenting, reviewing state-of-the-art MPC-based SCA frameworks, and discussing their benefits/drawbacks in practice. This seminar is valuable for students aiming for careers at the intersection of databases and security. Understanding how database queries are executed and optimized under secure multi-party computation is essential for building scalable, trustworthy collaborative data systems in an increasingly privacy-constrained world. See below a representative sample of the papers we'll study in the seminar: Bater et al., SMCQL: Secure Query Processing for Private Data Networks. Sohn et al., Alchemy: A Query Optimization Framework for Oblivious SQL. Liagouris et al., Secrecy: Secure collaborative analytics in untrusted clouds. • Bau et al., Orq: Complex Analytics on Private Data with Strong Security Guarantees