Knowledge Graph

Build reusable entity-relation graphs from research findings.

The knowledge-graph workflow structures findings into reusable relational memory.

How it works

  1. Parses findings and claims from your artifact/topic.
  2. Extracts entities and relation edges.
  3. Links claims to source-backed evidence.
  4. Produces a normalized graph artifact.

Real examples

/knowledge-graph outputs/scaling-laws-brief.md
bohr knowledge-graph outputs/scaling-laws-brief.md

Expected output

  • outputs/<slug>-knowledge-graph.md

Includes: entities, relations, and claim-link mappings.

Customization

  • Limit to a domain slice (for example, “only compute and data-quality concepts”)
  • Require high-confidence claims only
  • Include contradiction edges when available