ISSUE-87: Turtle file - SHACL vs RDFS vs OWL, or all?

Today we discussed how to proceed with the Turtle file and the data 
model describing the SHACL vocabulary itself. Some people argued that we 
should use RDFS, others suggested OWL, others would like to apply SHACL 
to itself.

I can certainly see use cases for all these options, and believe it 
would be quite doable to publish all these formats.

Let's look at an example (in Turtle):

SHACL ------------------

sh:AbstractResult
     a sh:ShapeClass ;
     rdfs:subClassOf rdfs:Resource ;
     rdfs:label "Abstract Result" ;
     rdfs:comment "Instances of subclasses of this class can be 
constructed during constraint validation." ;
     sh:abstract true ;
     sh:property [
         sh:predicate sh:subject ;
         sh:class rdfs:Resource ;
         sh:maxCount 1 ;
         rdfs:label "subject" ;
         rdfs:comment "The subject of triples involved in this result." ;
     ] ;
     ... many more sh:properties

OWL -------------------

sh:AbstractResult
     a owl:Class ;
     rdfs:subClassOf rdfs:Resource ;
     rdfs:label "Abstract Result" ;
     rdfs:comment "Instances of subclasses of this class can be 
constructed during constraint validation." ;
     rdfs:subClassOf [
         a owl:Restriction ;
         owl:onProperty sh:subject ;
         owl:allowedValues rdfs:Resource ;
     ] ;
     rdfs:subClassOf [
         a owl:Restriction ;
         owl:onProperty sh:subject ;
         owl:maxCardinality "1"^^xsd:nonNegativeInteger ;
     ] ;
     ...

sh:subject
     a owl:ObjectProperty ;
     rdfs:label "subject" ;
     rdfs:comment "The subject of triples involved in this result." .


I would argue that the OWL can be automatically generated from the SHACL 
file. A pure RDFS file would be very poor in terms of contents - 
basically just a collection of named resources with comments, ranges and 
domains. But if it helps some people, why not...

Those who argue for an OWL or RDFS version should suggest how it would 
look like, e.g. how the constraint types are defined.

Holger

Received on Friday, 30 October 2015 06:02:31 UTC