Benchmark
PII-SCOPE
How often does a detector find personal information in text and assign it the correct type?
Explore the original evidence
Explore the original evaluation, its methods, and reported findings.
Open evalCoverage
Protegrity’s evaluation contains 8,030 densely annotated rows from several domains. A single passage can contain many different types of personal information.
Methodology
Recall measures how many labeled entities a detector finds. Typed recall also requires the correct category. Precision captures false alarms. Systems support different entity categories, which affects comparisons on typed detection.