named-entity-extraction

Named Entity Extraction

A legal document processing system receives contracts in plain text. Each document needs tokenization, named-entity recognition (people, organizations, dates, monetary amounts), and structured extraction into a JSON schema that downstream contract-management software can ingest.

Pipeline

[ner_tokenize]
     |
     v
[ner_tag]
     |
     v
[ner_extract_entities]
     |
     v
[ner_link]

Workflow inputs: text

Workers

ExtractEntitiesWorker (task: ner_extract_entities)

• Writes entities, entityCount

LinkWorker (task: ner_link)

• Uses randomization
• Writes linkedEntities

TagWorker (task: ner_tag)

• Filters with predicates
• Writes taggedTokens

TokenizeWorker (task: ner_tokenize)

• Writes tokens, tokenCount

---

8 tests | Workflow: ner_named_entity_extraction | Timeout: 60s

See RUNNING.md for setup and usage.