Suggestions for the SHACL version

[HolgerKnublauch]

I am playing a bit with the SHACL file from 4.0 beta. Some suggestions:

• the node shapes seem to have sh:targetClass pointing at themselves. But in that case they should also be instance of rdfs:Class. And once this happens, the sh:targetClass triple can be removed altogether, see https://www.w3.org/TR/shacl/#implicit-targetClass

• this is particularly relevant as they are used as sh:class of the property shapes. If the rdf:type rdfs:Class exists in other graphs then this other graph should be owl:imported.

• or is the intent to only use node shapes? Then sh:node should be used instead of sh:class.

• is there a way to include (human-readable) sh:name values for the property shapes, maybe they can be derived from the rdfs:labels of the property paths from some other graph?

• please add rdf:type sh:PropertyShape to all property shapes. The triple is formally not required but makes processing easier.

[DanSmith]

Hi Holger, thank you very much for reviewing the DDI beta 4.0 release and for your insightful suggestions!
I've added this as a challenge for the upcoming hackathon at the European DDI conference in Chur the beginning of December.
ddi-developers/.github#5