Re: model theory for RDF/S

From: Dan Connolly <connolly@w3.org>
Subject: Re: model theory for RDF/S
Date: Fri, 28 Sep 2001 11:55:21 -0500

> "Peter F. Patel-Schneider" wrote:
> > 
> > From: Pat Hayes <phayes@ai.uwf.edu>
> > Subject: Re: model theory for RDF/S
> > Date: Fri, 28 Sep 2001 10:57:10 -0500
> [...]
> > > Ah, I had not thought of that in the DAML+OIL context.  Can you give
> > > a simple example?
> > 
> > Consider the following, where one allows ranges to be XML Schema datatypes.
> > (I hope that this is more-or-less valid N3.)
> > 
> > John age 0000025 .
> > 
> > Susan phone-number 0000025 .
> > 
> > age rdfs:range xsd:int .
> > 
> > phone-number rdfs:range xsd:string .
> > 
> > If there is a unique node for the literal 0000025, then what is the literal
> > value that it maps to?  Is it the string "0000025" or the integer 25?
> 
> It's the string "0000025" and the age line is goofy;
> it has to be written
> 
>  John age _:i25.
>  _:i25 rdf:value "0000025".
> 
> There are no integer literals in N3/n-triples syntax (yet, anyway).
>
> Even if we make up syntax for integer literals in n-triples
> syntax, that won't change RDF 1.0 syntax, which has no
> integer literal terms.

Then I think that all the examples of N3 and RDF where the literals are
supposed to be integers should be modified!  (See the N3 primer for several
of these.)   Technically, RDF 1.0 has no string literals either, of
course.  Interpreting literals as integers makes just as much sense as
interpreting them as unicode character sequences (or as interpreting them
as stars).

I thought that the idea here was to see how literals could be given a
reasonable meaning.  One way of proceeding is to require that the meaning of a
literal be completely carried in its lexical form.  This would break much
existing RDF, I think.  Another way of proceeding is to deduce the meaning
of a literal from the context, which, in RDFS, need not be lexically
adjacent to the literal itself.


> There may be integer literal values in the RDF model theory,
> but they're not directly expressible in RDF/xml 1.0 syntax;
> in RDF/xml 1.0 syntax, you can only say the equivalent of:
> 
>   the integer whose XML schema lexical representation is "0000025"
> 
> or rather:
> 
>   an integer whose XML schema lexical representation is "0000025"
> 
> where you can deduce that there is only one such integer
> from axioms about xsd:int and rdf:value.

Not in the new RDF model theory , you can't, nor in RDF 1.0.  Maybe you
could do it in a KIF axiomatization of RDF, augmented with axioms about
integers and sequences of numerals.


> Actually, the syntax above only says the equivalent of
> 
>   something with rdf:value (aka XML schema lexical representation)
> "0000025"
> 
> you can deduce that something is an integer, i.e.
>  _:i25 rdf:type xsd:int.
> 
> from the range line and the age line.
> 
> -- 
> Dan Connolly, W3C http://www.w3.org/People/Connolly/

peter

Received on Friday, 28 September 2001 15:37:36 UTC