Generalise diagrams

[nicholascar]

Internal comments like [1] and some external ones indicate it's not appropriate to assume all classes in PROF diagrams are OWL classes. They could be RDFS classes and perhaps should be OWL/RDFS non-specific altogether.

Suggested actions:

1. generalise PROF diagram keys to remove references to OWL/RDFS
2. better indicate that xsd:datatypes are not classes
3. consider E-R diagrams instead of class diagrams
4. compare with DCAT & other W3C specs for expectations
5. review all diagrams and their keys for consistency

[1] https://lists.w3.org/Archives/Public/public-dxwg-wg/2019Mar/0456.html

[nicholascar]

Diagram key moved to it's own figure in Conformance section: https://raw.githack.com/w3c/dxwg/prof-3PWD-candidate/prof/#diagramconventions

[nicholascar]

Diagram key merged into ED: https://w3c.github.io/dxwg/prof/#diagramconventions

[nicholascar]

@kcoyle has this issue been addressed to your satisfaction?

[kcoyle]

@nicholascar I'm not sure which of my comments was intended to be responded to here, so I'll say:

1. the confusing inclusion of elements in the legend that are not in the diagram is resolved
2. I wasn't sure how to read it. In RDF diagrams, properties are generally shown linking to values, not to the range of the value. Since RDF has both domains and ranges the links to classes are ambiguous - if this is intended to show the ranges of the objects of the properties that needs to be made clear. Also, it shows one value (token), which, if this is a class diagram, should be rdfs:literal.
3. I still argue that dct:format is not appropriate as a property on a node or URI that is not a physical item with a format.

[nicholascar]

@kcoyle thanks for the itemisation.

For 2: "properties are generally shown linking to values"...
Not in ontology diagrams! See AGRIF or the W3C's Organization Ontology or even ADMS. ADMS shows the property dcat:dataset with the TBox triple adms:AssetRepository dcat:dataset adms:Asset as per a UML class diagram with the
adms:AssetRepository box connecting via a dcat:dataset arrow to a adms:Asset. I've property diagramming like prof:hasResource within prof:Profile prof:hasResource prof:ResourceDescriptor in the same way. Yes, to range values.

Also for 2: "Also, it shows one value (token), which, if this is a class diagram, should be rdfs:literal."
xsd:token a owl:DatatypeProperty . and the key indicated that the rectangles are "XSD Datatype" objects so this is correct.

For 3: "I still argue that dct:format is not appropriate as a property on a node or URI that is not a physical item with a format."
This is a substantive point about the modelling, not the diagrams per se so can you raisw a separate issue for that?

[kcoyle]

@nicholascar ADMS is a UML diagram that shows classes and attributes. Either do an RDF graph diagram or a UML class diagram. This diagram is neither.

Yes, the key indicates that rectangles are literals, but that still doesn't explain why that is the only value shown. If "has token" -> xsd:token then "has artifact" -> xsd:anyURI, etc. It's the inconsistency here that makes the diagram inherently illogical.

Also, the use of dct:Standard here causes me problems. First it shows that the profile class is a subclass of the dct:Standard class. Then it says that the profile is a profile of an rdf:Resource that happens to be a dct:Standard. The profile is not a profile of the class dct:Standard, but of an instance of that class. So there is one box that is both a class and an instance, and I have trouble seeing it as both. If dct:Standard is a class, and not an instance of a class, then you can have prof:Profile (an instance of that class) that is a subclass of dct:Standard, and rdf:Resource, that is an instance of dct:Standard, and prof:Profile that is a profile of that rdf:Resource (and, also, can be a profile of a prof:Profile, which isn't shown here).

As for dct:format, there is a separate issue for that: #13 with a diagram at https://github.com/w3c/dxwg/issues/769#issuecomment-469287460. And probably discussed elsewhere as well.