Skip to content

Json ld

di# JSON-LD in FAIR²

What is JSON-LD?

JSON-LD (JavaScript Object Notation for Linked Data) is a lightweight linked data format used in FAIR² to represent machine-readable metadata. It allows datasets to be: - Interoperable – Ensures compatibility with Schema.org, ML Croissant, and FAIR principles. - AI-Ready – Enables seamless integration into machine learning workflows. - Linked Data-Compliant – Supports globally unique identifiers and structured relationships.

FAIR² uses JSON-LD to describe datasets in a way that both humans and machines can understand.

Why JSON-LD for FAIR²?

Compatible with ML Croissant – Works with AI dataset metadata standards.
Supports Schema.org – Ensures datasets are discoverable by search engines.
Enhances FAIR principles – Provides rich semantic metadata.
Machine-Actionable – Facilitates AI & ML dataset integration.

🚀 Basic JSON-LD Structure in FAIR²

A FAIR² dataset metadata file (fair2.json) follows this structure:

{
  "@context": "https://fair2.ai/spec/fair2_context",
  "@type": "Dataset",
  "name": "Example AI Dataset",
  "description": "A dataset for training AI models.",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "distribution": [
    {
      "@type": "DataDownload",
      "contentUrl": "https://example.com/data.csv",
      "encodingFormat": "text/csv"
    }
  ]
}

FAIR² JSON-LD Context (@context)

The @context defines how terms in the dataset metadata map to standardized vocabularies.

Example FAIR² Context

{
  "@context": {
    "schema": "https://schema.org/",
    "xsd": "http://www.w3.org/2001/XMLSchema#",
    "cr": "https://mlcommons.org/ns/",
    "fair2": "https://fair2.ai/ns/"
  }
}

How @context Works

  • schema → Maps to Schema.org properties (e.g., schema:name).
  • xsd → Ensures correct datatypes (e.g., xsd:string).
  • mlc → Supports ML Croissant metadata (e.g., cr:citeAs).
  • fair2 → Defines FAIR²-specific extensions.

FAIR² Metadata Schema in JSON-LD

FAIR² extends Schema.org and ML Croissant to describe AI-ready datasets.

Dataset Metadata Example

{
  "@context": [
    "https://fair2.ai/spec/fair2_context",
  ],
  "@type": "Dataset",
  "name": "FAIR² AI Dataset",
  "description": "A dataset prepared for machine learning workflows.",
  "author": {
    "@type": "Person",
    "name": "Dr. Jane Doe",
    "affiliation": {
      "@type": "Organization",
      "name": "AI Research Lab"
    }
  },
  "citation": "Doe, J. (2025). FAIR² Dataset for AI Research.",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "distribution": [
    {
      "@type": "DataDownload",
      "contentUrl": "https://example.com/data.zip",
      "encodingFormat": "application/zip"
    }
  ],
  "cr:citeAs": "Doe, J. FAIR² AI Dataset (2025)",
  "dct:conformsTo": "https://fair2.ai/spec/"
}

Why JSON-LD Matters for FAIR²

Enhances dataset discoverability using Schema.org. Ensures machine-actionable metadata for AI pipelines. Supports FAIR principles by enabling structured linked data. Facilitates interoperability with ML Croissant & SHACL validation.

Next Steps

Explore the FAIR² Schema to structure datasets correctly. Validate JSON-LD using SHACL & RDF tools. Learn about SHACL Validation for quality assurance.

Conversion of JSON-LD Schemas to Turtle Format

FAIR² maintains both JSON-LD and Turtle (TTL) representations of its schema and ontology files. During the continuous integration and deployment (CI/CD) process, these files are automatically converted and synchronized using the script located at ontologies/jsonld_to_turtle.py. This ensures consistency between formats and guarantees that both human-readable (Turtle) and machine-actionable (JSON-LD) versions are always up to date when new releases are published.