Close

LinkML for MIDAS Catalog

A formally-modeled, standards-based representation of the catalog metadata

Executive Summary

This page introduces the LinkML representation of the MIDAS Catalog metadata — a single formally-modeled schema from which a JSON Schema, a JSON-LD context, an OWL ontology, and a browsable reference are all generated. It is additive: a second, standards-based rendering of the same records the schema.org JSON-LD describes, aimed at the semantic-web and data-modeling community.

At a Glance
  • One authored model, many artifacts — a JSON Schema (validation), a JSON-LD context, an OWL ontology, and browsable class/slot docs, all generated from one LinkML source.
  • Additive — a second representation of the same records, not a replacement for the schema.org JSON-LD.
  • Validatable & ontology-mapped — a formal, machine-checkable view for the semantic-web / data-modeling community.
  • Persistently identified — the model and records resolve under w3id.org/midas-catalog/.
Two representations of the same catalog
MIDAS publishes its catalog metadata in two independent, standards-based representations of the same records. The schema.org JSON-LD representation is the web-facing, FAIR-harvestable form used for discovery (including Google Dataset Search) and general interoperability. This page introduces the LinkML representation — a formally-modeled, ontology-mapped, validatable form for the semantic-web and data-modeling community. The two are not versions of one another; they are two renderings of one catalog.

Why a LinkML model?

The MIDAS Catalog's metadata is described by a single LinkML model — a formal, community-standard schema authored once, from which every downstream artifact is generated: a JSON Schema for validation, a JSON-LD context, an OWL ontology, and a browsable reference. Modeling the catalog this way gives it a few things a hand-built JSON-LD emission alone does not:

  • A formal, inspectable schema in a language the metadata-modeling community already reads.
  • Validation — records can be checked against the generated JSON Schema.
  • Ontology mappings — fields and controlled-vocabulary axes are mapped to community ontologies where applicable, so the meaning of the data is explicit and machine-resolvable.
  • Typed construction — the model generates typed (Pydantic) classes, so a consumer can build and work with records programmatically rather than assembling raw JSON by hand.
  • A browsable reference — generated documentation for every class and slot.

The LinkML representation is additive: it does not replace the schema.org JSON-LD, and both describe the same records. It exists so the catalog is legible and reusable by the semantic-web and data-modeling community, and as an exemplar of a formally-modeled research-metadata catalog.

What you can do with it

  • Browse the model reference — every class and slot, with class diagrams: MIDAS Catalog LinkML reference
  • Validate records against the generated JSON Schema: …/linkml/schema.json
  • Construct records programmatically using typed (Pydantic) classes generated from the published model.
  • Consume the LinkML-native JSON-LD — the same catalog records, published in LinkML-native form alongside the schema.org JSON-LD in the MIDAS Catalog.
  • Reuse the model and its mappings — the LinkML source, JSON-LD context, and OWL ontology are published for reuse.
New to LinkML?
LinkML (the Linked data Modeling Language) is an open, community-standard way to author data models that generate JSON Schema, JSON-LD, OWL, documentation, and typed code from one source. You do not need to know LinkML to use the MIDAS Catalog metadata — the schema.org JSON-LD and the generated JSON Schema are usable on their own. The LinkML model is there for those who want a formally-modeled, ontology-mapped view of the catalog, or who work in the LinkML/semantic-web ecosystem.

How it's published

The model (the LinkML source and the generated JSON-LD context, JSON Schema, OWL ontology, and reference documentation) is published in the public midas-catalog-linkml repository and its GitHub Pages site. The records — the LinkML-native JSON-LD for each catalog entry — are published in the MIDAS Catalog, regenerated nightly alongside the schema.org JSON-LD. Every record and the model's artifacts are identified by persistent w3id.org/midas-catalog/ identifiers.

Which representation should I use?

Use the schema.org JSON-LD representation for web discovery and general interoperability — it is what search engines and FAIR harvesters consume. Use the LinkML representation when you want a formally validated, ontology-mapped, typed view of the same catalog, or when you are building on it in a LinkML/semantic-web context. For the schema.org representation and the catalog's FAIR assessment, see MIDAS Catalog JSON-LD Metadata Architecture.