FAIR² Ontology¶

Overview¶

The FAIR² Ontology defines the core semantic entities and relations used across FAIR² metadata and datasets. It provides a structured, machine-readable vocabulary that ensures interoperability between data, methods, and publications within the FAIR² ecosystem. The ontology reuses existing standards from Schema.org, PROV-O, and the Contributor Role Ontology (CRO), while extending them with specific terms for AI-ready data curation and Responsible AI alignment.

The ontology is expressed in both JSON-LD and Turtle formats and is compatible with SHACL validation shapes defined under the fair2s: namespace.

Namespaces¶

Prefix	Namespace URI
`schema:`	https://schema.org/
`prov:`	http://www.w3.org/ns/prov#
`cr:`	http://mlcommons.org/croissant/
`credit:`	https://credit.niso.org/contributor-roles/
`fair2:`	https://fair2.ai/ns/
`fair2s:`	https://fair2.ai/shapes/
`rdfs:`	http://www.w3.org/2000/01/rdf-schema#
`rdf:`	http://www.w3.org/1999/02/22-rdf-syntax-ns#
`xsd:`	http://www.w3.org/2001/XMLSchema#
`oa:`	http://www.w3.org/ns/oa#

Classes¶

The following classes define the core conceptual entities in FAIR². Each class extends a compatible Schema.org or PROV-O superclass to ensure semantic interoperability.

`@id`	`rdfs:label`	`rdfs:comment`	`rdfs:subClassOf`
`fair2:DataArticle`	DataArticle	Scholarly article describing and linking an open dataset with methods and reuse guidance.	`schema:ScholarlyArticle`
`fair2:DataPortal`	DataPortal	An authored, versioned web application presenting a single FAIR² dataset — an interactive explorer or landing site, not a catalogue and not a bare API endpoint.	`schema:CreativeWork`
`fair2:DataArchive`	DataArchive	A long-term archive or repository that preserves the dataset.	`schema:CreativeWork`
`fair2:Section`	Section	Structured methodology section grouping procedural steps.	`schema:HowToSection`
`fair2:MethodSection`	MethodSection	Legacy alias for `fair2:Section`; retained for backward compatibility.	`schema:HowToSection`
`fair2:Step`	Step	Atomic procedural instruction within a method.	`schema:HowToStep`
`fair2:Substep`	Substep	Subordinate procedural instruction nested under a Step.	`schema:HowToStep`
`fair2:AuthorRole`	AuthorRole	Role played by an author in producing the dataset (e.g. Corresponding Author).	`schema:Role`
`fair2:ContributorRole`	ContributorRole	Contributor role drawn from CRediT/CRO vocabularies.	`schema:Role`
`fair2:DescriptiveStatistics`	DescriptiveStatistics	Summary statistics computed over a RecordSet or Dataset.	`schema:Dataset`
`fair2:Submission`	Submission	Submission package that groups dataset, article, and supporting materials for review.	`schema:CreativeWork`
`fair2:Certification`	Certification	Pointer node asserting the FAIR² certification or validation status of a package, and referencing the external credential.	`schema:CreativeWork`
`fair2:StepCase`	StepCase	Conditional branch within a method, carrying the condition and the path taken when it holds. Sits in the same fair2:step list as a Step.	`schema:HowToStep`
`fair2:ComplianceLevel`	ComplianceLevel	A FAIR² compliance status asserted for a package.	`skos:Concept`
`fair2:CertificationScope`	CertificationScope	A facet of a package that a certification covers.	`skos:Concept`
`fair2:SummaryStatistic`	SummaryStatistic	A kind of descriptive statistic, named in `cr:dataType`.	`skos:Concept`
`fair2:FAIR2CertificationCredential`	FAIR2CertificationCredential	A W3C Verifiable Credential asserting a package's certification status.	`schema:CreativeWork`
`fair2:FAIR2CertifierRegistry`	FAIR2CertifierRegistry	The signed list of entities authorised to issue FAIR² certifications.	`schema:CreativeWork`

Properties¶

The FAIR² properties define relationships between datasets, methods, contributors, and derived artifacts. They support semantic traceability and alignment with FAIR data and Responsible AI requirements.

`@id`	`rdfs:label`	`rdfs:comment`
`fair2:attachment`	attachment	Points to a supporting file or supplemental material.
`fair2:bugFixes`	bugFixes	Bug-fix entries on a changelog `UpdateAction`.
`fair2:citationKey`	citationKey	Short citation key identifying the dataset.
`fair2:dataArchive`	dataArchive	Links a dataset to the archive that preserves it.
`fair2:dataArticle`	dataArticle	Links a dataset or submission to its associated scholarly data article.
`fair2:dataPortal`	dataPortal	Links a dataset to the portal that hosts and serves it.
`fair2:dataset`	dataset	References the dataset resource associated with the entity.
`fair2:domain`	domain	Subject-domain classification of the dataset (e.g. a Wikidata concept).
`fair2:generated`	generated	Links a method step to the RecordSet fields or artifacts it produced (lineage).
`fair2:improvements`	improvements	Improvement entries on a changelog `UpdateAction`.
`fair2:isExperimentRelated`	isExperimentRelated	Flags whether a field relates to experimental data.
`fair2:manuscript`	manuscript	References a manuscript file associated with the submission.
`fair2:methodSection`	methodSection	Links a dataset to its methodology section(s).
`fair2:newFeatures`	newFeatures	New-feature entries on a changelog `UpdateAction`.
`fair2:next`	next	Orders sequential method sections, steps, or substeps.
`fair2:nextTrue`	nextTrue	Conditional branch target taken when a StepCase condition holds.
`fair2:otherInformation`	otherInformation	Miscellaneous notes on a changelog `UpdateAction`.
`fair2:qualifiedUsage`	qualifiedUsage	Condition guarding a conditional method step (StepCase).
`fair2:statistics`	statistics	Links to computed descriptive statistics.
`fair2:step`	step	Connects a Section with its constituent Step items.
`fair2:store`	store	Indicates the storage location or repository endpoint for an asset.
`fair2:substep`	substep	Connects a Step with its constituent Substep items.
`fair2:unit`	unit	Unit of measurement for a field or variable.

Certification vocabulary¶

Two small controlled vocabularies. The compliance level says who vouches; the scope says what was checked — there is no metadata-only level.

Individual	Type	Meaning
`fair2:Certified`	`fair2:ComplianceLevel`	An authority assertion, backed by a credential signed by an authorised certifier.
`fair2:Validated`	`fair2:ComplianceLevel`	A quality status: the automated checks pass. Self-assertable, provided the claim carries its evidence.
`fair2:MetadataConformance`	`fair2:CertificationScope`	The metadata conforms to the declared FAIR² profile.
`fair2:DataIntegrity`	`fair2:CertificationScope`	The data files were retrieved and checked against their digests.
`fair2:LicenseVerification`	`fair2:CertificationScope`	The declared licence was checked.
`fair2:ProcessAttestation`	`fair2:CertificationScope`	The production process was attested by the certifier.
`fair2:TemporalProof`	`fair2:CertificationScope`	An RFC 3161 timestamp or blockchain anchor is attached.
`fair2:DistinctValueCount`	`fair2:SummaryStatistic`	Number of distinct values. DDI-CDI has no equivalent.
`fair2:ModalFrequency`	`fair2:SummaryStatistic`	How often the modal value occurs; DDI names the mode but not its frequency.
`fair2:ExpectedValueCount`	`fair2:SummaryStatistic`	Values the extraction expected to find — completeness, not distribution.
`fair2:FrequencyDistribution`	`fair2:SummaryStatistic`	Frequency of the most common values, as an `rdf:JSON` literal.

A certification whose scope omits dataIntegrity MUST be rendered as “FAIR²-Certified (metadata only)”. That is a requirement on consumers — SHACL can check the scope tokens, but not how they are displayed.

Added in v1.6.0 — status `testing`¶

First needed by a producer's output (the data wrangler) and defined where no term in the vocabularies a fair2.json already uses says the same thing. They carry vs:term_status "testing": usable, but open to change until a second producer relies on them or they settle. Where an existing vocabulary covers a case, a document uses it instead — e.g. csvw:null for missing-value codes, cr:equivalentProperty schema:latitude for a coordinate field, geocr:timeSeriesIndex for a time axis, oa:Annotation for an issue.

Describing the data¶

Term	Meaning
`fair2:role`	A field's role: `identifier`, `dimension`, `measure`, `count`, `category`, `datetime`, `geo`, `filemeta`. Identifiers correspond to `cr:key`; dimension, measure and attribute-like roles to the RDF Data Cube component properties.
`fair2:roleSource`	Whether that role was `asserted` or `computed`.
`fair2:valueConstraints`	The values a field may take, in words.
`fair2:unitSingular`	What one record is (`cell`, `patient`).
`fair2:unitPlural`	Its plural.
`fair2:unitLabelSource`	Where those labels came from.
`fair2:primaryMeasure`	The field an overview of a record set leads with.
`fair2:encrypted`	Whether a file's content is encrypted.
`fair2:rowCount`	Rows in a table.
`fair2:easting`	A projected easting coordinate — named by a field's `cr:equivalentProperty`.
`fair2:northing`	A projected northing coordinate, likewise.
`fair2:declaredLicenses`	Every licence a source declared, when it declared several.
`fair2:InferredJoin`	A join between record sets found by inference and not verified (unlike `cr:references`).
`fair2:CompositeKey`	A verified foreign key over several fields, when the target's `cr:key` is already another set of fields — otherwise Croissant's own form is used: a `cr:references` per field into the target's key.
`fair2:fromField`	The field(s) an inferred join or composite key starts from.
`fair2:toField`	The field(s) it ends at.
`fair2:cardinality`	`one-to-one`, `one-to-many`, `many-to-one` or `many-to-many`.
`fair2:confidence`	`high`, `medium` or `low`.
`fair2:matchedCount`	Distinct values or tuples found on both sides.
`fair2:sourceDistinct`	Distinct values or tuples on the starting side.
`fair2:targetDistinct`	Distinct values or tuples on the other side.
`fair2:matchRate`	The share matched, 0–1.
`fair2:uniqueness`	How unique the joined key is on the other side, 0–1.

Recording how the data was processed¶

Term	Meaning
`fair2:PrivacyScreening`	Activity assessing whether a dataset can be published without disclosing personal information.
`fair2:subjectType`	Who or what the data is about.
`fair2:decision`	The screening's outcome.
`fair2:action`	The change required of a field, on the `oa:Annotation` (motivated by `oa:editing`) that targets it.
`fair2:userOverride`	Whether the screening's decision was overridden.
`fair2:narrativeWithheld`	Whether narrative detail was withheld from the public record.
`fair2:IntegrityAssessment`	Activity assessing a dataset's consistency and completeness; its issues are `oa:Annotation`s motivated by `oa:assessing`.
`fair2:overallScore`	One score summarising it.
`fair2:checks`	The individual checks and results, as an `rdf:JSON` literal.
`fair2:severity`	How severe an issue is, on its annotation.
`fair2:complianceStatus`	How far an applied standard was complied with.
`fair2:contextLoad`	Whether a documentation file was preserved, and how.
`fair2:renameRule`	How a set of files was renamed when packaged, as an `rdf:JSON` literal.
`fair2:retrievedOn`	When a source was retrieved.
`fair2:retrievalMethod`	How it was obtained.
`fair2:resolvedBy`	How a dataset's main entity was resolved.
`fair2:matchSource`	Where a match to an ontology concept came from.

Governance — added in v1.7.0, status `testing`¶

The governance blocks a fair2.json may carry, from the FAIR² governance model. Each is optional and holds a JSON value (rdf:JSON): its contents are defined by schemas/fair2-governance.schema.json, not by vocabulary terms. Documents write the keys with their prefix. See Governance Blocks and the Governance Model.

Term	Meaning
`fair2:access`	The access tier and the conditions on access (modifiers +G, +S, +C, +P, +J, +Q). Absent, or without a tier, means L0.
`fair2:constraints`	Use constraints: AI training, export, commercial use, combination.
`fair2:embargo`	A date before which governed data is not released.
`fair2:variableGovernance`	Tiers and rules for columns, files, asset classes and record partitions.
`fair2:temporalGovernance`	Access that changes over time.
`fair2:agentGovernance`	Whether and how AI agents may use the data.
`fair2:consentGovernance`	How consent withdrawals affect access.
`fair2:indigenousGovernance`	Indigenous governance framework, bodies and cultural protocols.
`fair2:distributionGovernance`	How keys reach recipients outside a broker network.
`fair2:encryption`	How governed data is encrypted — written by a governance service.
`fair2:AccessTier`	Class of the three access tiers.
`fair2:L0`	Open.
`fair2:L1`	Registered.
`fair2:L2`	Credentialed.

fair2:constraints here is a dataset's use constraints. A field's value constraints are fair2:valueConstraints.

Deprecated in v1.7.0 — Agreement Levels¶

Retained only so existing references resolve; each names its closest tier with dct:isReplacedBy. See Migrating from Agreement Levels.

Level	Identifiers	Replaced by
0 — Embargoed	`fair2:AgreementLevel0`, `fair2:AgreementLevels0`	`fair2:L1` with an embargo
1 — Open	`fair2:AgreementLevel1`, `fair2:AgreementLevels1`	`fair2:L0`
2 — Collaborative	`fair2:AgreementLevel2`, `fair2:AgreementLevels2`	`fair2:L0`
3 — Trusted	`fair2:AgreementLevel3`, `fair2:AgreementLevels3`	`fair2:L2`
4 — Community-governed	`fair2:AgreementLevel4`, `fair2:AgreementLevels4`	`fair2:L1` with +G
5 — Secure environment	`fair2:AgreementLevel5`, `fair2:AgreementLevels5`	`fair2:L2` with +S
6 — Federated	`fair2:AgreementLevel6`, `fair2:AgreementLevels6`	`fair2:L2` with +S and +Q

Reserved terms¶

These four are declared so that existing documents keep resolving, but they have no shape, no example and no stated domain or range. Other unused terms have been removed; these are retained because the FAIR² specification has published them as reserved. They carry vs:term_status "unstable" in the ontology. Consumers should not rely on them, and producers should not emit them, until they are specified.

Term	Intended meaning
`fair2:Submission`	A submission package grouping dataset, article and supporting materials for review. Whether it ever appears in a published `fair2.json`, as opposed to a pre-publication review package, is not yet decided.
`fair2:manuscript`	Reference to a manuscript file in a submission package.
`fair2:attachment`	Reference to supplemental material in a submission package.
`fair2:store`	Storage location or repository endpoint for an asset. Its relationship to `schema:contentUrl` on a `FileObject` and to `fair2:dataArchive` is undefined.

Notes¶

Each property and class is compatible with JSON-LD serialization and SHACL validation.
The fair2s: namespace hosts corresponding SHACL shape definitions for data validation.
The ontology is designed for seamless integration with ML Schema, Croissant, and PROV-O provenance structures.