Re: More on XSD in RDF

[Patrick Stickler, Nokia/Finland, (+358 40) 801 9690, patrick.stickler@nokia.com]

> If there is anything to add to the Concepts WD, maybe it would be to
> spell out the exact requirements for datatypes used in RDF
> datatyping, some of which is in the introduction to the above.  Maybe
> near http://www.w3.org/TR/rdf-concepts/#section-Datatypes ?

I'm wondering if it is not too strong a requirement that datatyping
entailments involving relations between datatypes must be based
on explicit rdfs:subClassOf assertions -- to emphasize that the
entailments are based on the intersection of their value spaces;
and it remains then a "social" issue whether the subclass assertions
should/could be made if the definitions of the datatypes themselves
do not define such relations.

It is straightforward to capture the official subclass relations
defined by the XML Schema specs, and those assertions would not
include e.g. a relationship between xsd:float and xsd:decimal,
therefore it would be clear insofar as RDF is concerned that
the following entailment does not hold, according to the XML Schema
specs:

{
   thing:A some:Property "10.0"^^xsd:decimal .
   thing:B some:Property "10.0"^^xsd:float .
}
does not datatype entail
{
   thing:A some:Property _:x .
   thing:B some:Property _:x .
}

And if folks want to then challenge the fact that the entailment
does not hold, they can go rag on the XML Schema folks, but RDF
is fully capturing and respecting the formal definitions of those
datatypes and their non-relation.

And it's by rdfs:subClassOf that we make the basis for such
entailments explicit in RDF.

Patrick

Received on Monday, 25 November 2002 05:02:33 UTC