RE: Justification for new node type

> -----Original Message-----
> From: ext Graham Klyne [mailto:Graham.Klyne@MIMEsweeper.com]
> Sent: 11 August, 2002 16:32
> To: Stickler Patrick (NRC/Tampere)
> Cc: w3c-rdfcore-wg@w3.org
> Subject: Re: Justification for new node type
> 
> 
> At 07:05 PM 8/9/02 +0300, Patrick.Stickler@nokia.com wrote:
> >I would like the proponents of the recent proposal for
> >a new datatyped literal node type to justify why URIs
> >cannot be used.
> 
> I believe there is a strong technical reason why URIs don't work 
> here.  According to the model theory 
> (http://www.w3.org/TR/rdf-mt/#sinterp), URIs can denote 
> anything according 
> to the interpretation that is used.  There is no provision 
> for treating 
> different classes of URIrefs differently in this respect.

Well, I wouldn't go so far as to consider that a license
to override any semantics imposed by the URI scheme itself.
If a URI scheme is formally defined to denote members of
a particular class of resource, then the RDF MT does not
override that to say that such URIs can be used to denote
any arbitrary resource.

Rather, the MT simply says (or should say) that anything can
be denoted by *a* URI, but what that particular URI is, and
what URI scheme is used to denote it, is a completely
disjunct issue. Eh?

Granted, one may certainly assert some interpretation that
conflicts with RDF-external constraints (and certainly
URI scheme imposed constraints are external to RDF) but
that doesn't mean that -- taking the whole enchilada -- that
such an interpretation is acceptable at all levels.

I think there is an implicit but necessarily expectation that
all URIs dealt with by RDF are "correct and valid" both
syntactically and semantically, so one could say that assigning
an interpretation to a URI which conflicts with the semantics
of that URI (via its URI scheme) is incorrect.

> Conversely (and as Jan pointed out long time ago - 
> http://ioctl.org/rdf/literals) a literal is a fixed value, 
> not subject to 
> reinterpretation according to the "possible world" being described.
> 
> (It might be possible to redraw the model theory to use URIs 
> in the way you 
> suggest, but I think that would be a far greater disruption 
> than what is 
> being proposed.)

Well, Jan has already pointed out a very valid and practical
reason why URIs wouldn't work as a general solution, based
on magnitude constraints, so I guess we can just leave this
one alone ;-)

Cheers,

Patrick

Received on Monday, 12 August 2002 03:35:59 UTC