ko-pii
A Python library for detecting and pseudonymizing personal information in Korean documents
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
ko-pii detects personal information in Korean documents using rules, dictionaries, and checksums without external ML dependencies, then pseudonymizes it reversibly. It blocks PII during RAG ingestion and retrieval and preserves context by replacing the same person with the same token. It also supports permission-based restoration through a Vault and audit trails.
KEY FEATURES
Key features
- Detect Korean personal information and pseudonymize it reversibly
- Block PII during RAG ingestion and retrieval
- Preserve context with consistent tokens for the same person
- Support permission-based restoration and audit trails
EDITOR'S NOTE
Why did it stand out?
The workflow is specific: it blocks personal information at both ingestion and retrieval, while replacing the same person with the same token to preserve context. It also keeps human review in the loop for false positives instead of applying learned changes automatically.