Re: support XML Literals in RDF

From: Graham Klyne <GK@ninebynine.org>
Subject: Re: support XML Literals in RDF
Date: Tue, 12 Aug 2003 12:26:31 +0100

> At 05:27 08/08/03 -0400, Peter F. Patel-Schneider wrote:
> > > > Although any RDF-only application, i.e., an application that *only*
> > > > needs to determine the RDF implications of an RDF graph, does not need
> > > > any special code to support XML Literals beyond the code needed to
> > > > support sequences of octets, an RDF application that goes beyond these
> > > > implications, for example to determine whether a literal is in LV, will
> > > > need considerable code to support XML Literals.
> >
> >I was wrong.
> >
> >Even rdf-entailment requires access to an oracle to determine whether an
> >octet-sequence is in canonical form.  Rule rdf2 requires access to such an
> >oracle.
> 
> I don't think that anyone has claimed this is not the case ... my 
> understanding has always been that a C14N oracle or something is needed to 
> do *full* RDF entailment.  But for many useful applications that use RDF, 
> such full entailment may not be required.
> 
> #g

Perhaps, but other formalisms built on top of RDF (such as OWL) may be more
dependant on entailment.  My understanding of the situation is that one
(perhaps unstated) reason for the changes was to allow a reasoner for a
formalism built on top of RDF (such as an OWL) to not require an oracle for
XML canonical form.  This has not been achieved.

Peter F. Patel-Schneider
Bell Labs Research
Lucent Technologies

Received on Tuesday, 12 August 2003 08:06:43 UTC