Enhancing Privacy through an Interactive On-demand Incremental Information
summary
Research builds an interactive record-linkage interface that reveals PII only on-demand incremental disclosure, logs every access, and enforces a hard budget. A controlled experiment shows utility (accuracy and completion time) stays the same while disclosure falls from 100% to 7.85% privacy risk (static methods earlier hit ~30%). Expert reviews report partial disclosure can improve decisions versus full dump; the system targets the legal minimum necessary principle and defends against insider misuse.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:52Data Science Data wrangling
- 2:29Automating Machine Learning
- 2:53Legitimate Access Without Informed Consent
- 3:58Partial Solutions
- 5:27Differential Privacy
- 6:09Privacy by Design
- 8:22Key Design Elements
- 10:13Background
- 11:11Privacy Goals
- 11:29The Sweet Spot
- 12:40Results
- 13:53Dynamics Interface
- 14:54Main Results
- 15:51Expert Study
- 16:40Threat Models
- 17:08The Team
- 17:56Audience Questions
related talks
32mPrivate Insights into AI Use
15mEnhancing Privacy through the Visual Design of Privacy Notices: Exploring the Interplay
28mPrivacy Engineering Networking Break AMA
20mShare data without revealing personal information
40mApplied Privacy Engineering Study
39mPrivacy-preserving Data Sharing
28mHow Privacy Engineers Deliver ROI
01h 00mStepping-up: The Census Bureau Sets an Example of How to Be a Good Data Steward in the 21st Century