Re: Formation of RDF terms

Jonathan,

I pretty much agree with most of your response.

Maybe where we differ is that I would like to see the model straightened 
out, or clarified, THEN sort out the syntactical oddities with reference to 
a stable, well-understood model.

To me, the model isn't just a topic of theoretical discussion:  it has 
intrinsic utility.  The syntax is needed for exchanging model data, for 
embedding model data in XML documents and (in some cases) for providing an 
interpretation of existing XML documents in terms of model 
constructs.  (Not using the terms "model" and "interpretation" here in a 
logician's sense.)

#g

At 02:23 PM 1/29/01 -0500, Jonathan Borden wrote:
>Graham Klyne wrote:
>
> > At 07:55 AM 1/29/01 -0500, Jonathan Borden wrote:
> > >so clearly the 'concept' or term 'unsignedInt' has a URI assigned by the
>XML
> > >Schema spec of:  http://www.w3.org/2000/10/XMLSchema#unsignedInt.
> > >
> > >IMHO, it is not reasonable to have RDF and (e.g.) XML Schema assign a
> > >different URI to the same thing.
> >
> > I agree it's not reasonable, but I don't think it's invalid.
>
>I respectfully suggest that such inconsistencies (e.g. between RDL and XML
>Schema), especially when the RDF URI looks either weird or unreasonable, are
>not helpful in improving acceptance of RDF among the wider web community.
>
> >
> > I think there are three difficult areas:
> >
> > (1) RDF-syntax is not internally consistent with these things:  in some
> > cases ID values are simply concatenated with some URI;  in other cases
>they
> > are treated as fragment identifiers attached to a URI.
>
>Assuming RDF-syntax desires to be internally consistent with itself and with
>RDF Schema syntax, I presume that the intention of the attributes "ID" and
>"rdf:ID" is to define an XML ID .... which is a fragment identifier.
>
>I've always thought one of the beauties of a typedNode is that the rdf:type
>resolves to the actual Node in the RDFS document where the Class is defined.
>This of course involves a base URI + fragment identifier.
>
>Whether we like it or not RDF is heavily dependent on RFC 2396 and the
>characteristics of a URI ... with warts and all.
>
> >
> > (2) As you note, RDF-syntax practice doesn't sit comfortably with related
> > XML practices.
> >
> > (3) As noted elsewhere, recently, RDF treatment of resources differs from
> > Web treatment of resources -- an RDF resource is not necessarily a web
> > resource.
>
>Ouch. There have been many many systems for representing directed graphs
>using triples and quads. Perhaps the only thing unique and interesting about
>RDF is the use of URIs and all that they imply. Loose that and why are we
>bothering to have this discussion?
>
> >
> > Like Dave, I happen to think it's the model that really matters, but
>having
> > these syntactic oddities sometimes makes it more difficult to discuss the
> > model clearly.  (And I think point (3) is a model-related problem.)
> >
>
>Aside from theoretical discussions of the RDF model, if it isn't helpful in
>practice, then who cares? To me the use of RDF is to discuss web
>resources -- otherwise I've got a whole bunch of database systems and SQL
>queries lying around -- as well as a whole bunch of theory books describing
>normalization etc. If what you say is really true (I'm unconvinced that this
>is not all syntactic.), then fixing this discrepancy is more important than
>purity of whatever (IMHO)
>
>-Jonathan

------------
Graham Klyne
(GK@ACM.ORG)

Received on Wednesday, 31 January 2001 13:10:31 UTC