Governed growth: keeping a continuously ingested, provenance-first knowledge graph trustworthy

A governance method for knowledge graphs that ground AI answers and grow daily from sources of very different quality. Nine principles — source authority tiers, identity before enrichment, no synthesised identifiers, rule-classed relations, schema re-validation, licence-carrying data, reviewed judgement, observed growth and minimised personal data — on a deterministic, idempotent pipeline, mapped to PROV-O, SHACL, ELI, ECLI, LEI and the EU AI Act. Method only; no implementation, parameters or graph contents.