Knowledge graphs

The knowledge graph is in development. This page describes the approach.

Most AI tools search through chunks of text. FNDRBLDR stores what the text actually claims: entities, claims and relationships, each linked to the evidence behind it.

Three layers

• Ontology: the types and relationships a domain allows, versioned and approved by a person.
• Knowledge: the entities, claims and hypotheses themselves.
• Evidence and provenance: source documents, exact passages, dates and how each was verified.

Rules it follows

• A hypothesis never silently becomes a fact.
• Contradictions are kept side by side, not overwritten.
• Copies of one article count as one source.
• You can ask what was known at any past date.
• Agreement between AI models is not evidence.

Your data stays portable

Graphs will export with their provenance to JSON triples, JSON-LD and RDF, among other formats.