Re: rdfs:subClassOf and metaclasses

At 20:11 30/06/2002 -0700, Piotr Kaminski wrote:

>Hi,
>
>In short:
>
>Do the subject and object of an rdfs:subClassOf statement have to both be
>instances of the same (meta)class?

They must both be instances of rdfs:Class, though this can be inferred 
rather than specified.  There is no other constraint.  I don't think this 
answers your question though.


>In long:
>
>Are the following RDF triples valid (modulo syntax errors):
>
>MA rdf:type rdfs:Class
>MA rdfs:subClassOf rdfs:Class
>MB rdf:type rdfs:Class
>MB rdfs:subClassOf rdfs:Class
>A rdf:type MA
>B rdf:type MB
>A rdfs:subClassOf B

There is nothing wrong with this set of triples, though they do look a 
little strange.  It might best to explore what you think you are trying to 
say.  What do you think the subClassOf triples are for?

The following is also correct RDF:

   A rdfs:subClassOf B

A schema aware RDF processor may conclude from this:

A rdf:type rdfs:Class
B rdf:type rdfs:Class


>The only thing I can find is in the RDFS spec is section 2.3.2, which says
>that "Only instances of rdfs:Class can have the rdfs:subClassOf property and
>the property value is always of rdf:type rdfs:Class".  The statements above
>seem to satisfy this constraint.

Yes they do, but the say more than is necessary to specify a simple 
subClassOf relationship.  The minimum that is needed is a single triple:

   Human rdfs:subClassOf Animal .


>The only related RDFCore WG issue I can find is rdfs-clarify-subClass-and
>instance.
>
>Incidentally, does anyone know how UML deals with this?

This is not the right place to ask UML questions.

Brian

Received on Monday, 1 July 2002 03:07:42 UTC