Look before you Link: Privacy Risk Inspection of Open Data through...
summary
A visual analytic workflow called pry clusters open datasets using weighted clustering, generates all pairwise joins (a small cluster of seven datasets yields about 28 combinations), and ranks those pairs by a computed risk score so analysts focus on the riskiest joins. It auto-suggests join attributes, lets users inspect matched records and flips a vulnerable record switch that revealed one exact GPS-tagged arrest for a 76-year-old white female, demonstrating practical identity disclosure.
AI-written summary · about
related talks
22mEnhancing Privacy through an Interactive On-demand Incremental Information
15mThe Missing Link in Privacy Risk Assessments
32mThe Dangerous Data Anonymization - Keynote
43mAnalyzing Administrative Data with Privacy Protection in Place
22mPrivacy Threat Modeling
41mMatching Uses and Data Protections for Government Data
20mPrivacy Shift Left: A Machine-Assisted Threat Modeling Approach
40mProtecting sensitive data in huge datasets #bbuzz