VocPub Profile - Specification

1.1 Data Expansion

Some SKOS elements - classes and predicates - can be inferred based on rules present in the SKOS model. For example, skos:broader and skos:narrower are inverse predicates thus if I have <A> skos:broader <B>, I can infer <B> skos:narrower <A>.

This profile allows data to be supplied in a minimalist form that is not conformant to this specification until a series of calculations, based on certain SKOS rules, are carried out on it through a process known as data expansion.

The particular rules that may be applied to data before validation with this profile's validator are given in the table below. These rules are enacted by the application of a series of SPARQL^ref queries to the data, all of which are packaged up inside a SHACL^ref file in this profile's repository.

CONSTRUCT {
    $this skos:hasTopConcept ?c .
}
WHERE {
    ?c skos:topConceptOf $this .
}

CONSTRUCT {
    $this skos:topConceptOf ?cs .
}
WHERE {
    ?cs skos:hasTopConcept $this .
}

CONSTRUCT {
    $this skos:broader ?n .
}
WHERE {
    ?n skos:narrower $this .
}

CONSTRUCT {
    $this skos:narrower ?b .
}
WHERE {
    ?b skos:broader $this .
}

CONSTRUCT {
    $this skos:inScheme ?cs .
}
WHERE {
    $this skos:broader*/skos:topConceptOf ?cs .
}

CONSTRUCT {
    $this ?p ?o3
}
WHERE {
    $this skos:inScheme ?cs .

    VALUES ?p {
        prov:wasDerivedFrom
        skos:historyNote
        sdo:citation
        dcterms:source
        dcterms:provenance
    }

    ?cs ?p ?o .

    OPTIONAL {
        $this ?p ?o2 .
    }

    BIND (COALESCE(?o2, ?o) AS ?o3)
}

Application of these rules will allow a Concept supplied without any provenance predicates to be calculated from its containing Concept Scheme, thus satisfying Requirement 2.3.4.

The SHACL file containing all the expansion queries is available at:

• https://w3id.org/profile/vocpub/expander

A Python script able to execute the expansion rule on RDF data before validation is available at:

• https://w3id.org/profile/vocpub/validate

1.2 Dependencies

To characterise vocabularies according to the mandatory and suggested Requirements of this Profile, several other vocabularies will need to be used. This profile is therefore dependent on those vocabularies. For this reason, copies of these vocabularies are maintained within the repository of this profile and are accessible as per the details table below.

2.1 Vocabulary

This profile identifies Semantic Web objects with URI-based persistent identifiers. For this reason:

§ 2.1.1 Each vocabulary MUST be identified by a IRI

As per the SKOS Primer^ref, a document guiding the use of SKOS:

concepts usually come in carefully compiled vocabularies, such as thesauri or classification schemes. SKOS offers the means of representing such KOSs using the skos:ConceptScheme class.

For this reason, this profile requires that:

§ 2.1.2 Each vocabulary MUST be presented as a single Concept Scheme object

For ease of data management:

§ 2.1.3 Each vocabulary MUST be presented in a single RDF file which does not contain information other than that which is directly part of the vocabulary

To ensure vocabularies can be catalogued effectively and governed:

§ 2.1.4 Each vocabulary MUST have exactly one title and at least one definition indicated using the skos:prefLabel and the skos:definition predicates respectively that must give textual literal values. Only one definition per language is allowed

NOTE: Unlike the general directions for the use of SKOS (the SKOS "Primer") labels in multiple languages should be indicated with skos:altLabel predicates, not all with skos:prefLabel, i.e. there should only ever be one skos:prefLabel value. If multiple definitions are given, the one in the language of the label is considered primary.

§ 2.1.5 Each vocabulary MUST have exactly one created date and exactly one modified date indicated using the sdo:dateCreated and sdo:dateModified or dcterms:created and dcterms:modified predicates respectively that must be either an xsd:date, xsd:dateTime or xsd:dateTimeStamp literal value

§ 2.1.6 Each vocabulary MUST have at least one creator, indicated using sdo:creator or dcterms:creator predicate and exactly one publisher, indicated using sdo:publisher or dcterms:publisher, all of which MUST be IRIs indicating instances of sdo:Person, or sdo:Organization. A prov:qualifiedAttribution predicate indicating an Agent with the prov:hadRole predicate indicating the value isorole:originator or isorole:publisher may be used instead of sdo:creator & sdo:publisher, respectively

To be able to link SKOS vocabularies to their non-vocabulary source information:

§ 2.1.7 The origins of a Concept Scheme MUST be indicated by at least one of the following predicates: skos:historyNote, sdo:citation, prov:wasDerivedFrom. dcterms:source MAY be used instead of sdo:citation and dcterms:provenance MAY be used instead of skos:historyNote but the schema.org and SKOS predicates are preferred.

If a vocabulary is based on another Semantic Web resource, such as an ontology or another vocabulary, prov:wasDerivedFrom should be used to indicate that resource's IRI. If the vocabulary is based on a resource that is identified by a IRI but which is not a Semantic Web resource, sdo:citation should be used to indicate the resource's IRI with the xsd:anyURI datatype. If the vocabulary is based on something which cannot be identified by IRI, a statement about the vocabulary's origins should be given in a literal value indicated with the skos:historyNote predicate. If the vocabulary is not based on any other resource or source of information, i.e. this vocabulary is its only expression, this should be communicated by use of the skos:historyNote indicating the phrase "This vocabulary is expressed for the first time here".

The use of dcterms:source & dcterms:provenance is to maintain compatability with previous versions of VocPub only and may eventually be disallowed.

To ensure that all the terms within a vocabulary are linked to the main vocabulary object, the Concept Scheme:

§ 2.1.8 All Concept instances within a Concept Scheme MUST be contained in a single term hierarchy using skos:hasTopConcept / skos:topConceptOf predicates indicating the broadest Concepts in the vocabulary and then skos:broader and/or skos:narrower predicates for all non-broadest Concepts in a hierarchy that contains no cycles.

To unambiguously link the term hierarchy within a vocabulary to the vocabulary itself:

§ 2.1.9 Each vocabulary's Concept Scheme MUST link to at least one Concept within the vocabulary with the skos:hasTopConcept predicate

To communicate the Registry Status of the vocabulary:

§ 2.1.10 The status of the vocabulary as a whole, according to the Registry Status standard^ref, SHOULD be given with the predicate reg:status indicating a Concept from the Registry Statuses vocabulary (https://linked.data.gov.au/def/reg-statuses).

To indicate whether and if so how this vocabulary has been derived from another vocabulary:

§ 2.1.11 The derivation status of the vocabulary SHOULD be given should be given with the predicate prov:qualifiedDerivation indicating a Blank Node that contains the predicated prov:entity, to indicate the vocabulary derived from and prov:hadRole to indicate the mode of derivation which SHOULD be taken from the Vocabulary Derivation Modes vocabulary (https://linked.data.gov.au/def/vocdermods).

Example data for a vocabulary indicating that it is an extension to another vocabulary using the mechanism defined in Requirement 2.1.2 is:

# Vocab X is derived from Vocab Y and is an extension of it
<http://example.com/vocab/x>
    a skos:ConceptScheme ;
    skos:prefLabel "Vocab X"@en ;
    ...
    prov:qualifiedDerivation [
        prov:entity <http://example.com/vocab/y> ;  # Vocab Y
        prov:hadRole <https://linked.data.gov.au/def/vocdermods/extension> ;
    ] ;
.

To high-level theming of a vocabulary:

§ 2.1.12 High-level theming of a vocabulary SHOULD be given using the sdo:keywords predicate indicating Concepts from another vocabulary. Alternatively, dcat:theme MAY be used. Text literal values for either predicate SHOULD NOT be used.

To indicate license, copyright:

§ 2.1.13 Any licence pertaining to the reuse of a vocabulary's content SHOULD be given using the sdo:license predicate preferentially indicating the IRI of a license if in RDF form or a literal URL (datatype xsd:anyURI) if online but not in RDF form. If the licence is expressed in test, a literal text field may be indicated.

§ 2.1.14 The copyright holder for the vocabulary SHOULD be given using the sdo:copyrightHolder predicate preferentially indicating the IRI of an Agent or a Blank Node instance of an Agent containing details as per Agent requirements. A prov:qualifiedAttribution predicate indicating an Agent with the prov:hadRole predicate indicating the value isorole:rightsHolder may be used instead of sdo:copyrightHolder.

2.2 Collection

From the SKOS Primer^ref:

SKOS makes it possible to define meaningful groupings or "collections" of concepts. Collections may contain Concepts defined in any vocabulary, not just the one the Collection itself is defined in.

To ensure that Collection instances are identifiable and their meaning isn't obscure or lost:

§ 2.2.1 Each Collection MUST have exactly one title and at least one definition indicated using the skos:prefLabel and the skos:definition predicates respectively that must give textual literal values. Only one definition per language is allowed

NOTE: Unlike the general directions for the use of SKOS (the SKOS "Primer") labels in multiple languages should be indicated with skos:altLabel predicates, not all with skos:prefLabel, i.e. there should only ever be one skos:prefLabel value. If multiple definitions are given, the one in the language of the label is considered primary.

If a Collection's grouping of Concepts is derived from an existing resource that is different from the ConceptScheme it is defined within:

§ 2.2.2 The origins of a Collection, if different from its containing Concept Scheme, SHOULD be indicated by at least one of the following predicates: skos:historyNote, sdo:citation, prov:wasDerivedFrom. dcterms:source MAY be used instead of sdo:citation and dcterms:provenance MAY be used instead of skos:historyNote but the schema.org and SKOS predicates are preferred.

For compatability with previous versions of this Specification, dcterms:provenance MAY be used instead of skos:historyNote but the latter is the preferred predicate.

To help list Collections within vocabularies:

§ 2.2.3 A Collection exists within a vocabulary SHOULD indicate that it is within the vocabulary by use of the skos:inScheme predicate. If it is defined for the first time in the vocabulary, it should also indicate this with the rdfs:isDefinedBy predicate

To ensure that a Collection isn't empty:

§ 2.2.4 A Collection MUST indicate at least one Concept instance that is within the collection with use of the skos:member predicate. The Concept need not be defined by the Concept Scheme that defines the Collection

2.3 Concept

From the SKOS Primer^ref:

The fundamental element of the SKOS vocabulary is the concept. Concepts are the units of thought — ideas, meanings, or (categories of) objects and events—which underlie many knowledge organization systems

Vocabularies conforming to this profile must present at least one Concept within the vocabulary file and, as per requirements in Section 2.1, at least once Concept must be indicated as the top concept of the vocabulary.

To ensure that Concept instances are identifiable and their meaning isn't obscure or lost:

§ 2.3.1 Each Concept MUST have exactly one title and at least one definition indicated using the skos:prefLabel and the skos:definition predicates respectively that must give textual literal values. Only one definition per language is allowed

NOTE: Unlike the general directions for the use of SKOS (the SKOS "Primer") labels in multiple languages should be indicated with skos:altLabel predicates, not all with skos:prefLabel, i.e. there should only ever be one skos:prefLabel value. If multiple definitions are given, the one in the language of the label is considered primary.

To ensure that every Concept is linked to the vocabulary that defines it:

§ 2.3.2 Each Concept in a vocabulary MAY indicate the vocabulary that defines it by use of the rdfs:isDefinedBy predicate indicating a Concept Scheme instance. If no such predicate is given, the Concept Scheme in the file that a Concept is provided in is understood to be the defining Concept Scheme

Note that the vocabulary that defines a Concept does not have to be the vocabulary in the file being validated. This is to allow for Concept instance reuse across multiple vocabularies.

Since a Concept may be used in more than one vocabulary:

§ 2.3.3 Each Concept in a vocabulary MUST indicate that it appears within that vocabulary's hierarchy of Concepts either directly by use of the skos:topConceptOf predicate indicating the vocabulary or indirectly by use of one or more skos:broader / skos:narrower predicates placing the Concept within a chain of other Concepts, the top concept of which uses the skos:topConceptOf predicate to indicate the vocabulary.

If a Concept is derived from an existing resource and that derivation is not already covered by source information for the vocabulary that it is within:

§ 2.3.4 The origins of a Concept, if different from its containing Concept Scheme, SHOULD be indicated by at least one of the following predicates: skos:historyNote, sdo:citation, dcterms:source or prov:wasDerivedFrom or dcterms:provenance.

If a Concept is based on another Semantic Web resource, such as another Concept or other defined object, prov:wasDerivedFrom should be used to indicate that resource's IRI. If the Concept is based on a resource that is identified by a IRI but which is not a Semantic Web resource, dcterms:source should be used to indicate the resource's IRI. If the vocabulary is based on something which cannot be identified by IRI, a statement about the vocabulary's origins should be given in a literal value indicated with the skos:historyNote predicate. If the vocabulary is not based on any other resource or source of information, i.e. this vocabulary is its only expression, this should be communicated by use of the skos:historyNote indicating the phrase "This vocabulary is expressed for the first time here".