FAIR² Example Walkthrough (Updated)¶

This walkthrough illustrates how the FAIR² ontology and metadata model are applied in a real dataset. The examples below are drawn directly from the Borja et al. (2025) FAIR² dataset (borja2025.json) and demonstrate how constrained shapes and entities interlink across submissions, data articles, methods, contributors, and results.

All examples are truncated to 1–2 representative instances for clarity.

Data Article¶

The FAIR² Data Article describes the dataset and its scientific context.

{
  "@id": "https://doi.org/10.3389/focsu.2024.1528837",
  "@type": "fair2:DataArticle",
  "schema:name": "Structure-Based Prediction of SARS-CoV-2 Variant Properties Using Machine Learning on Mutational Neighborhoods",
  "schema:publisher": {
    "@type": "schema:Organization",
    "schema:name": "Frontiers in Ocean Sustainability"
  },
  "schema:author": [
    {
      "@type": "schema:Person",
      "schema:name": "Borja, Efren"
    },
    {
      "@type": "schema:Person",
      "schema:name": "Schultes, Erik"
    }
  ]
}

This JSON-LD fragment illustrates how a fair2:DataArticle aligns with schema:ScholarlyArticle while referencing real contributors.

Method¶

Methods in FAIR² are structured hierarchically as fair2:Section objects containing ordered fair2:Step and fair2:Substep items.

{
  "@id": "https://fair2.ai/examples/borja2025#method",
  "@type": "fair2:Section",
  "schema:name": "Machine Learning Workflow",
  "schema:description": "End-to-end workflow from Bio2Byte feature extraction to model evaluation.",
  "fair2:step": [
    {
      "@type": "fair2:Step",
      "schema:name": "Feature Extraction",
      "schema:description": "Bio2Byte descriptors were extracted for each protein variant.",
      "fair2:step": [
        {
          "@type": "fair2:Substep",
          "schema:name": "Normalization",
          "schema:description": "Features were normalized using z-scores."
        }
      ]
    }
  ]
}

This structure demonstrates how the FAIR² Method representation supports multi-level procedural hierarchy and captures computational steps with clarity.

Contributors and Roles¶

Contributor information in FAIR² uses schema:Contribution entities enriched with role identifiers from the Contributor Role Ontology and FAIR² extensions.

{
  "@type": "schema:Contribution",
  "schema:agent": {
    "@type": "schema:Person",
    "schema:name": "Efren Borja",
    "schema:identifier": {
      "@type": "PropertyValue",
      "propertyID": "ORCID",
      "value": "https://orcid.org/0000-0002-1825-0097"
    },
    "schema:affiliation": {
      "@type": "schema:Organization",
      "schema:name": "Frontiers"
    }
  },
  "prov:hadRole": [
    {
      "@id": "https://credit.niso.org/contributor-roles/data-curation/",
      "rdfs:label": "Data Curation"
    },
    {
      "@id": "https://credit.niso.org/contributor-roles/formal-analysis/",
      "rdfs:label": "Formal Analysis"
    }
  ]
}

This example shows how FAIR² contributions maintain explicit attribution through PROV-compatible prov:hadRole relationships.

RecordSet and Statistics¶

FAIR² datasets include RecordSets linked to computed statistics and variable definitions.

{
  "@id": "https://fair2.ai/examples/borja2025#recordset",
  "@type": "cr:RecordSet",
  "schema:name": "RBD Variant Features",
  "field": [
    {
      "@id": "https://fair2.ai/examples/borja2025#field",
      "@type": "cr:Field",
      "description": "Variant name",
      "dataType": "schema:Text",
      "fair2:statistics": {
        ...
      }
    }
  ]
}

This section exemplifies how FAIR² captures analytical data products, linking them to their computed statistics and variable-level metadata.

Visualization¶

Access conditions are stated by the dataset's governance blocks. This dataset is open, so it states the open tier, L0, in fair2:access, and names the same tier in accessRights for consumers that read Dataset metadata alone:

"accessRights": {
  "@id": "https://fair2.ai/ns/L0",
  "@type": "DefinedTerm",
  "name": "L0 — Open",
  "definition": "No authentication is required. The data is distributed unencrypted; the licence and use terms apply, but do not gate access.",
  "url": "https://fair2.ai/spec/v1.7.0/specification/governance/#access-tiers"
},
"fair2:access": { "tier": "L0" }

A governed dataset adds the conditions and use constraints in the same blocks: see the complete example.

Summary¶

These examples illustrate the implementation of the FAIR² ontology in a real dataset. The JSON-LD structures demonstrate how methods, data, contributors, and derived outputs are interconnected through consistent FAIR² linking properties, enabling full traceability and AI-ready reuse.