De-identification in practice: professionals as end users
summary
Deidentification often fails in practice: small sets of nonunique attributes can reidentify people, k-anonymity fails on realistic assumptions, and the US Census defers differential privacy for some 2025 uses. Practitioners struggle with trial‑and‑error toolchains (a 2015 educational dataset required many configurations) and need clearer guidance. The project performs qualitative coding of guides plus interviews, surveys and user studies to uncover concrete pain points and produce better guidance, open‑source tooling, or evidence for adopting differential privacy debated approaches.
AI-written summary · about
related talks
39mPrivacy-preserving Data Sharing
01h 09mDynamic data obscurity.
40mData Privacy and Security by Design
39mData Privacy Engineering to power Responsible Enterprise AI
50mPrivacy in Practice with Smart Pseudonymization: Lessons from the Belgian Public Sector
43mPanel: Privacy Design Patterns for AI Systems: Threats and Protections
08mPrivacy by Design for DevOps
52mDeveloping privacy