Protegrity PII model detects personal data fast with high accuracy
Protegrity-PII Technical Report
Computation and LanguageArtificial Intelligence
Summary
Personal Identifiable Information (PII) like names or addresses can be tricky to find in all kinds of text. The authors developed Protegrity-PII, a tool that spots PII in both clear and messy texts very accurately. It works fast, scanning a million characters in under half a second using a standard GPU. Because it was trained with many types of PII, it can recognize new or unexpected PII better than some other tools.
What this means in practice
- •For data privacy teams: Automatically identify and protect personal data in mixed text formats with improved accuracy and speed for compliance purposes.
- •For cloud service providers: Integrate fast, low-cost PII detection into large-scale text processing pipelines to enhance data security offerings.$Commercial implications: Enables new or improved automated data protection services for clients using scalable, affordable PII detection technology.
Authors
Hardeep Bassi, Hui Ma, Luís Seabra, Koowum Joshi, Daniel Pereira, Averell Gatton
Abstract
We present Protegrity-PII, a new PII decision model for structured and unstructured text. Protegrity-PII achieves a typed recall of 0.915 with a precision of 0.907 on a new benchmark we propose, PII-SCOPE. Our model processes text in 0.49 seconds per million characters at \$0.0005 per million characters on a single RTX 6000 using batched inference. Due in part to the expansive nature of the entity type taxonomy used in training, Protegrity-PII achieves superior out-of-distribution generalization when benchmarked against a selection of state of the art models and datasets.