# NLP Overview

# NLP Overview

Natural Language Processing (NLP) in Curiosity Workspace turns raw text into structured signals you can search, filter, and connect to your graph.

You typically use NLP to:

• extract entities and concepts from text
• normalize content across languages and writing styles
• link mentions in text to existing nodes in your graph
• improve retrieval and downstream AI workflows

# How NLP fits into the platform

NLP interacts with:

• Graph: entities can become nodes; links become edges (mentions → entities).
• Search: extracted fields can be indexed and used as facets.
• AI: LLM workflows become more reliable when grounded on extracted and linked entities.

# Key building blocks

• Pipelines
  • sequence of steps applied to a field (tokenization, entity detection, etc.)
• Models
  • spotters/patterns/classifiers used by pipelines to capture entities
• Entity capture
  • the act of extracting entities from text into structured outputs
• Entity linking
  • connecting captured entities to existing nodes (or creating new ones when appropriate)

# When to use NLP (and when not to)

Use NLP when:

• your critical information is embedded in free text (tickets, notes, transcripts)
• you need structured facets that don’t exist as explicit fields
• you want graph navigation from text mentions

Avoid overusing NLP when:

• your source already provides structured fields (use connector mapping first)
• entity capture would be too noisy without domain tuning

# Next steps

• Configure semantic retrieval: Embeddings
• Extract structured signals: Entity Extraction