Required / Recommended properties of supporting classes

[jakubklimek]

Problem statement
For certain supporting classes such as Category Scheme (skos:ConceptScheme), GeoDCAT-AP 2.0.0 lists some properties as recommended, or even mandatory. For instance, title (dct:title) is mandatory for Category Scheme.

This would mean that all category schemes used in GeoDCAT-AP records always need to have a title, even if they are represented by their IRIs, e.g.:

<catalogue> a dcat:Catalog ;
   dcat:themeTaxonomy <http://publications.europa.eu/resource/dataset/eurovoc>,
                      <http://publications.europa.eu/resource/dataset/data-theme> .

<http://publications.europa.eu/resource/dataset/eurovoc> dct:title "EuroVoc"@en .
<http://publications.europa.eu/resource/dataset/data-theme> dct:title "Data theme"@en .

as presence of mandatory properties should be validated.
Even if this was left to dereference of the ConceptScheme IRI, we should not impose on ConceptScheme creators to use dct:title if they want to be used with GeoDCAT-AP. For instance, this is the case of the EU Data theme NAL, which does not have a dct:title.

In addition, this causes problems in a cross-profile environment, as whenever GeoDCAT-AP would be part of the used profile mix, all Concept Schemes would have to have the mandatory title (dct:title), or errors would be produced.

A slightly ligher problem is the one of recommended properties. For instance, label (rdfs:label) is Recommended for a Media Type. However, members of the IANA Media Types code-list should be used e.g. for dcat:mediaType or dcat:packageFormat. This would mean recommending adding a label to those items, e.g.

[] a dcat:Distribution ;
  dcat:mediaType <http://www.iana.org/assignments/media-types/application/xml> .
<http://www.iana.org/assignments/media-types/application/xml> rdfs:label "XML"@en .

which does not seem desirable nor should it be recommended.

Proposal
For supporting classes such as Category Scheme, Standard, Kind, etc. the listed properties should be all optional, which also removes the need to call them optional.

[jakubklimek]

Voting on this issue postponed to give the community more time to discuss it here in GitHub.