BIBO namespace document

[tombaker]

In 2026, the Usage Board will create new properties and classes in the BIBO Ontology namespace (http://purl.org/ontology/bibo/) in order to fill gaps in a Scholarly Resources Application Profile (SRAP) (see issue tracker for the "dc-srap" repository).

BIBO is currently modeled in what I think of as a "Protégé" style of ontology design that emerged in the early years of OWL (mid-2000s). In this style, properties from other namespaces that may be needed in metadata about bibliographic resources are in effect re-declared in the BIBO ontology document. Examples include foaf:name, dcterms:publisher, and prism:doi which are declared in the BIBO ontology document to be of type owl:ObjectProperty or owl:DatatypeProperty. In my understanding, this style strove to ensure that an ontology was self-contained, in effect as a one-stop model for resource description. I characterize this as a "Protégé" style because, if I recall, the tool itself would added some axioms automatically, for example to ensure that the ontology fit the model of description logic (OWL-DL).

The DCMI style of modeling, in contrast, assumes that properties and classes are declared (by their owners) in namespace documents, then reused in "application profiles". Modern "shape languages" such as ShEx and SHACL build on this distinction.

I'm wondering if BIBO could or should now be expressed in the DCMI style with:

1. A minimal namespace document and schema, strictly for declaring properties and classes in the BIBO namespace, on the model of DCMI Metadata Terms. For reasons of consistency and backwards compatibility, these terms could continue to be defined as owl:DatatypeProperty and owl:ObjectProperty.
2. Optionally, a BIBO application profile, expressed perhaps in a DCTAP document, which would use all properties and classes of the current BIBO ontology from non-BIBO namespaces, such as foaf:name. Axioms in the current BIBO ontology that make assertions about non-BIBO terms could also be expressed there, for example the skos:scopeNote on dcterms:issued which says that the property is "Used to describe the issue date of a bibliographic resource".

The resulting namespace document would declare BIBO properties and classes in a more focused and maintainable representation.

[tombaker]

This process could also split out the namespaces for document statuses (http://purl.org/ontology/bibo/status/) and for thesis degrees (http://purl.org/ontology/bibo/degrees/). Currently, the thesis degrees are modeled as follows:

bibo:degree rdfs:range bibo:ThesisDegree .
bibo:ThesisDegree rdf:isDefinedBy "http://purl.org/ontology/bibo/"^^<http://www.w3.org/2001/XMLSchema#anyURI> .
bibo:ThesisDegree rdf:type owl:Class .
bibodegrees:ma rdf:type owl:Thing .
bibodegrees:ma rdf:type owl:NamedIndividual .
bibodegrees:ma rdf:type bibo:ThesisDegree .
bibodegrees:ms rdf:type owl:Thing .
bibodegrees:ms rdf:type owl:NamedIndividual .
bibodegrees:ms rdf:type bibo:ThesisDegree .
bibodegrees:phd rdf:type owl:Thing .
bibodegrees:phd rdf:type owl:NamedIndividual .
bibodegrees:phd rdf:type bibo:ThesisDegree .

Might these bibodegrees: URIs be more reusable in application profiles if modeled as concepts in SKOS concept schemes?

[tombaker]

One minimally disruptive approach could be to leave the current BIBO ontology document untouched but to create, in parallel, "BIBO Vocabulary" as a minimal namespace document.

The scope of that namespace document would be most of the axioms about things in the BIBO namespace - ie, things with URIs that have the BIBO namespace URI (http://purl.org/ontology/bibo/) as their base URI and not including URIs that have the other base URIs, such as http://purl.org/ontology/bibo/status/ and http://purl.org/ontology/bibo/degrees/.

The namespace document for the BIBO Vocabulary, then, would be where the Usage Board would add the new properties and classes currently being approved. The PURL for the BIBO namespace would resolve to that namespace document.

Nothing in the new namespace document would contradict anything in the ontology document, with its richer set of axioms about things in multiple namespaces and OWL restrictions, so I see no obvious harm for existing implementers of the BIBO Ontology in its current form.

BIBO does not yet fall explicitly under the DCMI Namespace Policy. It would be easier for DCMI to extend its commitment for semantic stability to BIBO if the object of that commitment were more tightly scoped to the BIBO namespace.

[tombaker]

Note that BIBO and DC Terms differ in their use of rdfs:isDefinedBy - see #4 for further discussion.