Knowledge Graphs

Knowledge graphs extract entities (people, organizations, locations) and relationships from your documents. They build a connected network that spans your entire document collection.

What graphs do

When you create a graph from documents, Moongraph:

1. Extracts entities — Identifies people, organizations, locations, dates, and other things mentioned in text
2. Finds relationships — Identifies how entities connect (works_for, located_in, etc.)
3. Resolves duplicates — Merges different mentions of the same entity ("J. Smith" and "John Smith")
4. Builds a network — Creates a connected graph you can explore and query

Getting started

Key features

Graph detail view

Click any graph to see:

• Overview — Entity and relationship counts, processing status, top connected entities
• Source Data — Documents included in the graph
• Entities — Paginated table of all extracted entities
• Relationships — Paginated table of all relationships
• Schema — Entity and relationship types used for extraction
• Activity — Processing history

Entity resolution

The same entity often appears differently across documents. Entity resolution automatically merges duplicates. When it misses connections, you can merge entities manually.

Visualization

Open graphs in Cosmograph for full-screen network visualization, or in Workbench for combined exploration and AI chat.

Sharing

Graphs are shared through folders. Add a graph to a folder, share the folder with your team, and team members can view and query the graph. See Share a Graph.

Learn more