PII Shield: Multilingual Personally Identifiable Information Detection Dataset

PII Shield is a large-scale, multilingual dataset for training and evaluating Personally Identifiable Information detection models. Built by Auren Research, it combines real-world documents from diverse domains with high-quality span-level PII annotations produced by fastino/gliner2-privacy-filter-PII-multi. The dataset achieved the highest F1 score on the SPY benchmark among open-source PII detectors.

Use Cases

• Train PII detection models based on multilingual text and span-level annotations.
• Benchmark PII detection performance based on the SPY benchmark metric.
• Evaluate model robustness on real-world documents from diverse domains mentioned in the description.

Strengths

• Large-scale dataset designed for multilingual PII detection.
• Annotations produced by fastino/gliner2-privacy-filter-PII-multi, which achieved the highest F1 on the SPY benchmark.
• Combines real-world documents from diverse domains.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2026-05-17 00:04:43; freshness should be verified.

Provenance

Source

Auren Research

Collection Method

Combines real-world documents with annotations from fastino/gliner2-privacy-filter-PII-multi.

Freshness

Last updated 2026-05-17 00:04:43.