Knowledge graphs, in plain language
Short, precise definitions of the terms used across our products and documentation.
- Knowledge graph
- Information stored as things and the connections between them, rather than as rows or documents.
- Ontology / vocabulary
- The agreed definitions for a domain: what a Supplier is, what an Ingredient is, and which may relate to which.
- Triple
- One fact in three parts: Supplier X — is located in — France.
- RDF / OWL / Turtle
- W3C standards for writing facts and vocabularies down so any tool can read them.
- SHACL
- A W3C standard for rules a graph must obey — the automated correctness check.
- Provenance
- The paper trail: which document, which passage, which run produced a given fact.
- LPG
- Labelled property graph — the model used by Neo4j, ArangoDB and similar engines.
- GSTL / contract
- GraFlo’s model file: one document describing the whole graph model, versioned like code.
- RAG / GraphRAG
- Answering questions by retrieving supporting material first — from text or from a graph — so the AI cites rather than invents.
- MCP
- The standard protocol AI agents use to call tools; GraFlo’s context plane speaks it natively.
Want to see these ideas at work on your data?