Re: Typed literals: current status

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


----- Original Message ----- 
From: "ext Jeremy Carroll" <jjc@hpl.hp.com>
To: <w3c-rdfcore-wg@w3.org>
Sent: 22 October, 2002 21:30
Subject: Re: Typed literals: current status


> 
> Jeremy (the deconstructionist):
> > aaa ppp <ddd>lll .
> > 
> > rdfs-entails
> > 
> > <ddd> rdf:type rdfs:Datatype .
> 
> 
> DanC (unreformed modernist?):
> > Gads, I hope not!
> 
> Thinking about it, what we were saying was that at the RDF level we were 
> treating the datatype-literal thing as a pair (or triple); its's only 
> entailment that knows that <ddd> is a datatype that makes anything more of a 
> <ddd>lll; this would weaken the entailment to be:
> 
> *empty*
> 
>  {ddd a datatype}-entails
> 
> <ddd> rdf:type rdfs:Datatype .
> 
> that looks a bit better to me.

Or rather, the act of putting a class URI <ddd> into a typed literal
embodies the implicit assertion that <ddd> rdf:type rdfs:Datatype,
because that is what typed literals contain, a *datatype* URI.

If that URI doesn't denote a datatype, then that is a semantic error
just as much as if the lexical form specified is not a member of the
lexical space of a datatype.

The term 'typed literal' really perhaps ought to be 'datatyped literal'
which it was originally but got shortened for convenience.

The original closure rule suggested by Jeremy is correct. Putting
a URI into a typed literal label asserts that it denotes a datatype
and applications should be allowed to assume that it does. Period.

Patrick

Received on Wednesday, 23 October 2002 01:25:31 UTC