AI-Assisted RDF Authoring from JSON Data

Abstract

Scientific workflows increasingly produce structured JSON data that is easy to exchange but hard to interpret consistently across systems: JSON Schema validates structure, but carries no Linked Data semantics. We present an RDF Authoring View that extends the open-source JSON Schema editor MetaConfigurator into an integrated, AI-assisted environment for turning existing JSON, YAML, XML, or CSV data into RDF within a single web interface. Researchers generate RML mappings with AI assistance, refine the resulting triples, edit JSON-LD @context definitions with ontology-aware IRI auto-completion, run SPARQL queries drafted from natural-language hints, and explore the data as an interactive knowledge graph. Across these steps, the AI drafts artifacts while the researcher stays in control through an explicit human-in-the-loop review. We demonstrate the workflow on laboratory data from metal-organic framework (MOF) synthesis, reuse established vocabularies aligned with the NFDI4Chem ecosystem, and validate the generated RDF against a SHACL shape.