[topicmapmail] Meaning of URIs - ongoing debate on new W3C forum

[Jan Algermissen]

"Thomas B. Passin" wrote:
> 
> Jan Algermissen wrote:
> > "Thomas B. Passin" wrote:
> >
> >>Jan Algermissen wrote:
> >>
> >>>Thomas, please answer this:
> >>>
> >>>a) in RDF, what does "http://www.w3.org/index.html" denote?
> >>>
> >>>
> >>>
> >>>b) in TM-land, what does "http://www.w3.org/index.html" denote?
> >>>
> >>
> >>I have no idea.
> >
> >
> > Well, 'denote' was the wrong word to use. I meant 'identify'.
> >
> > So in a) the URI per definition allways identfies the Web-Resource
> > (whatever that resource is intended to be in this case)
> 
> No it does not - the rfc for URIs specifically says that URIs can
> identify non-retrievable resources, including concepts.  

Sure. I did not say something different. I used the term Web-Resource to
refer to the 'Resource' term the Web architecure uses. Indeed, a Web-Resource
can identify anything (though I find this idea extremely questionable...but
that's how it is defined anyhow..)


RDF uses URIs
> as abstract "identifiers" when they appear as values of rdf:about or rdf:ID.
> 
> >  and in
> > b) we cannot be sure if the resource either identifies the WebResource or
> > a subject that is indicated by this Web-resource. We'd need the addressing
> > context to answer this.
> >
> > What I wanted to stress is that addressing in RDF and Topic Maps is really
> > different and not compatible.
> >
> 
> The use of a URI as a PSI in topic maps - I am still claiming - is very
> similar to the way that rdf uses a URI as the value of rdf:about or
> rdf:ID.  In both cases, the URI is being used to "identify" some concept

Again, in Web-land I allways know from e.g. http://www.w3.org/index.html
that it identifies a Web-Resource. Not so in topic maps. You cannot mix-in
the issue that Web architecture allows resources to be anything. That is
not the issue here.

> (which may or may not happen to be concrete, and may or may not happen
> to be web-retrievable).  It is true the the __psi__ is supposed to be
> retrievable, but not necessarily the concept that the psi describes.

So, what if the URI of a PSI is actually denoting an abstract concept?
doesn;t that cause confusion?

> 
> I have been putting quotes around "identify" because I do not mean to
> get into a discussion about technicalities and philosophy of "identity"
> and "identification".
> 
> > Furthermore, Topic Maps implicitly make the assumption that a URI identifies
> > a resource that 'is' a document. TMs completely ignore the possibility that
> > http://www.w3.org/index.html might actually identify 'the W3C'. A TM author
> > could really make the W3C an occurrence of some topic in that case.
> >
> 
> But that document is  not necesssarily the subject of a topic - it may
> only "indicate" the actual subject.  In effect, this use of a URI is
> two-fold -

Yes! And that is what I am trying to make clear: in Web-land the use of URI is 
**never** two fold, in Topic Maps it allways is.

> 
> 1) As a unique identifying string, and
> 2) As a pointer to a human-readable document that attempts to say
> somethng useful about the nature of the subject in question.
> 
> RDF does not require 2), but many prominent RDF people advocate
> following it as a convention.
> 
> Nothing in the RDF specs requires the string
> 'http://www.w3.org/index.html' to identify 'the W3C'.  If used as the
> value of an rdf:about or rdf:ID attribute, a string conforming to the
> syntax of a URI identifies a node in an RDF graph, not a node on the
> internet and not anything outside of the RDF graph.  The subject of that
> node (to use topic map terminology, but it is appropriate here) may or
> may not be the W3C.

Sure. But in any case the URI identifies a Resource.

> 
> Now, a particular RDF graph _could also_ use that URI to represent the
> W3C, but that would have to be done through some predicate or
> convention.  There appears to be some confusion about this in parts of
> the RDF community as well, to judge by Pat Haye's recent posts on the
> subject.  And it is certainly true that the language in the RDF specs -
> especially the original Model & Syntax Recommendation - can lead to some
> confusion about what is meant by the word "resource".

Well, the (to me crazy) idea is that the authority (only! the authority)
that controlls the namespace of an URI defines what the resource identified by
that URI denotes. In TM-land the TM-author makes that decision by using either
subject address of subjectIndicator.

> 
> The last word has not been said, I think, about the interpretation of an
> RDF resource when the identifying URI of that resource also happens to
> be dereferenceable.  But however that eventually shakes out, a URI used
> as the value of rdf:about or rdf:ID refers to a node in an RDF graph
> (whether or not it should be taken to also point to the "subject" of
> that node, if there is one).
> 
> The claim I am making is that the use of a URI as a subject indicator in
> topic maps is very similar to the above. 

Well then, try to convert some RDF into a Topic Map.....


> True, a psi is supposed to
> point to a document that a person can read, 

The RDF/TM interoperability problem comes in when a TM-author uses
for example http://www.w3.org/index.html to be a PSI for 'the W3C'
but the W3C does not define http://www.w3.org/index.html to denote
a document but 'the W3C'. When trying to bridge between RDF and
TMs you end up with the W3c indicating the W3C.
I am sure there are better examples.


Jan

> but if it did not actually
> do so, the processing of a topic map would not be affected.  In effect,
> the psi functions, as I said above, as an abstract string that serves to
> identify a topic in a topic map graph - st least if we believe that
> topics having the same subject should be (or could be) merged because
> they are really about one and the same subject.
> 
> Cheers,
> 
> Tom P
> 
> _______________________________________________
> topicmapmail mailing list
> topicmapmail@infoloom.com
> http://www.infoloom.com/mailman/listinfo/topicmapmail

-- 
Jan Algermissen                           http://www.topicmapping.com
Consultant & Programmer	                  http://www.gooseworks.org