Describing properties and classes in BIBO

[tombaker]

BIBO is an ontology modeled in what one might call a "Protégé" style of ontology design in which, for the sake of one-stop completeness, any required properties and classes defined in other namespaces such as DC, SKOS, and FOAF are copied into the ontology document and asserted to be instances of OWL classes such as owl:DatatypeProperty, owl:ObjectProperty, or owl:AnnotationProperty. See 'bibo' repo issue 3 for more discussion.

The Usage Board is currently deciding to add properties and classes to the BIBO namespace, http://purl.org/ontology/bibo/. Whether these new terms will be added to the current ontology document or to a new BIBO namespace document is the topic of 'bibo' repo issue 3.

This issue enumerates the properties of existing BIBO terms for discussion. We need to decide which properties to use for new terms in order to ensure that new terms are described in a manner consistent with existing BIBO classes (59), datatype properties (37), and object properties (30).

[tombaker]

BIBO and DC Terms both use the following:

• rdfs:label for the human-readable "label"
• rdfs:comment for the definition
• rdfs:domain and rdfs:range for domain and range, as needed.
• rdfs:subClassOf or rdfs:subPropertyOf, as needed.

BIBO and DC Terms differ in their use of rdfs:isDefinedBy:

• In BIBO, rdfs:isDefinedBy takes a literal object of type xsd:anyURI. Example:

    <http://purl.org/ontology/bibo/numPages> <http://www.w3.org/2000/01/rdf-schema#isDefinedBy> "http://purl.org/ontology/bibo/"^^<http://www.w3.org/2001/XMLSchema#anyURI> .

• In DC Terms, rdfs:isDefinedBy takes the namespace URI of DC Terms as object. Example:

    <http://purl.org/dc/terms/format> <http://www.w3.org/2000/01/rdf-schema#isDefinedBy> <http://purl.org/dc/terms/> .

BIBO and DC Terms differ in their basic typology of terms:

• BIBO classes are of type owl:Class, while DC Terms classes are of type rdfs:Class.
• BIBO properties are of type owl:DatatypeProperty or owl:ObjectProperty, while DC Terms properties are of type rdf:Property.

BIBO and DC Terms differ in how they express scope notes:

• DC Terms expresses a scope note with dcterms:description. Example (from dcterms:license): "Recommended best practice is to identify the license document...".
• There are no scope notes for existing BIBO properties and classes, though several are annotated with skos:editorialNote, skos:historyNote, or skos:example.

DC Terms uses dcterms:issued to record the date on which a term was first published. BIBO records no such dates.

[tombaker]

PROPOSAL

For new terms in BIBO, I propose that we use:

• rdfs:label for the human-readable "label"
• rdfs:comment for the definition
• rdf:type with owl:Class, owl:DatatypeProperty, or owl:ObjectProperty, as appropriate.
• rdfs:domain and rdfs:range for domain and range, as needed.
• rdfs:subClassOf or rdfs:subPropertyOf, as needed.
• rdfs:isDefinedBy with object "http://purl.org/ontology/bibo/"^^<http://www.w3.org/2001/XMLSchema#anyURI> subject possibly to future change as discussed in 'bibo' issues 3 and 4.
• dcterms:description with a scope note, if needed.
• I propose that we start using dcterms:issued for the date on which the BIBO term was first published, starting with the new properties and classes.

[kcoyle]

Does this mean that you are proposing to use owl:Class in the place of owl:AnnotationProperty? And note that there are some instances of owl:AnnotationProperty in BIBO but not in the BIBO namespace (cf. dcterms:title)

rdf:type with owl:Class, owl:DatatypeProperty, or owl:ObjectProperty, as appropriate.

[niklasl]

Regarding the annotation properties, note that (per the The OWL primer):

Annotation properties are used to encode information about (parts of) the ontology itself (like the author and creation date of an axiom) instead of the domain of interest.

and (further down in section 8.3):

OWL 2 DL requires that a name is not used for more than one property type (object, datatype or annotation property) nor can an IRI denote both a class and a datatype.

(Which is more restrictive than RDF itself, where everything (including literal values) is a resource, and rdfs:Datatype is a subclass of rdfs:Class. Also, in OWL 2 Full (the RDF-based semantics), datatype properties are object properties.)

I tend to think that it is problematic to include axioms about terms from other vocabularies like this in ontology documents, and have viewed the owl:AnnotationProperty usage in BIBO as Protégé artefacts that should possibly be removed (among some other things). But I don't use Protégé (and mainly work using OWL Full reasoning when applicable) so I don't know of any adverse practical effects if we remove it.

It is true that the following ones are used on ontology terms in BIBO, so for OWL DL these assertions might be needed (although perhaps Protégé could handle that internally; we could investigate a bit more).

annotation term	count	example subject term
sw:term_status	116	most if not all
owl:deprecated	1	bibo:content
skos:scopeNote	27	e.g. foaf:Person
skos:editorialNote	3	e.g. dct:Agent
skos:example	2	bibo:section
skos:historyNote	1	bibo:content
skos:changeNote	1	event:agent
skos:note	0

(Note that owl:deprecated a owl:AnnotationProperty is already in OWL, so it is just redundant.)

The DC terms dct:title, dct:description, and dct:creator are only used to describe the ontology itself, not on terms/axioms thereof; and dct:identifier isn't used on any resource here. So the AnnotationProperty assertions of these could possibly be removed with no adverse effects even for OWL DL usage.

[annakasprzik]

(searched a while for the namespace sw -- should that be ns?)

[niklasl]

(searched a while for the namespace sw -- should that be ns?)

It's for http://www.w3.org/2003/06/sw-vocab-status/ns (sw per https://prefix.cc/sw).