Applications · Established · Intermediate

Named Entity Recognition

Also known as: NER

Identifying and labelling spans of text that refer to people, organisations, places, dates, amounts or domain-specific entities.

What Named Entity Recognition is

NER turns prose into structured records, which is the first step in most document automation and knowledge graph pipelines.

How it works

Traditionally a sequence-labelling model over token spans; increasingly a language model returning a structured schema, which handles custom entity types with no training data.

Why it matters

Contract review, medical coding, invoice processing and compliance redaction all depend on reliable entity extraction.

Common uses

→Contract and invoice data capture
→PII detection and redaction
→Clinical information extraction
→Knowledge graph construction

Strengths

✓Directly produces structured data
✓Custom entity types via prompting

Watch for

✓Ambiguous boundaries
✓Domain drift hurts accuracy

Continue exploring

Structured Output Knowledge Graph Document AI Data Privacy in AI