Fact Extraction via Canonical Identifiers: Validating a Multi-Registry Pipeline for Dutch Open Data

Paper 4 in the Prioris research series. Validates fact-level edge extraction using seven Dutch canonical identifier types (BIG, ECLI, KvK, SKJ, QID, BRIN, AGB). Reports precision and recall per identifier type, compares regex vs spaCy NER, and demonstrates cross-registry verification for accountability use cases. v2: 10 issues fixed from Codex CLI review. Part of the Evidence-Graded Knowledge Graphs research programme (Paper 0: 10.5281/zenodo.19024611).