RE: web proper names

Looking at: 
http://www.hackcraft.net/rep/rep.xml

Jon's take seems interesting, but on a second-cut I would, in the
style of XTM and WPNs, add a class of "referents" or "subjects"
to denote resources that are primarily used to denote something
without a web-accessible representation - since after all, 
representations on the web are resources too. So it isn't covered by 
the case of "http://example.net/rep#NoRepresentation". Now, some files 
such as Expanded WPNS (WPNs over http://) are actually meant to be
used to refer to a "thing/referent" but have a representation.

Also, as regards Passim's predicates:
1) subjectIsTheThingReturnedByThisURI
2) theDocumentAtThisUriDescribesTheSubject
3) theDocumentAtThisUriIsAboutTheSubject

I would replace "Document" with "Representation", and I'm not sure
about this "About" versus "Describes" distinction. Seems unclear.
Is "About" meant that the RDF subject URI is a rigid designator-esque
URI meaning the "thing/referent", not the "representation". Can you
clarify? 

It's because of these issues I think a common "representation" format
such as WPN would be easier, and an Expanded Web Proper Name (or 
something from a expanded-Jon "referent" class) could then connect those 
predicates to a URI that has representations (information resource) over 
http://. My reason for choosing the Expanded Web Proper Name 
representation format would be that it allows easy "eye-balling" of the
referent\thing and includes relevant information (search engine terms and 
results) to find out if what two people "things/referents" URI are 
actually the same - in fact, since it's lists of URIs a machine might
have a heuristic as well. 

I think we should go ahead on all fronts, new RDF predicates and Web 
Proper Names. I would like to see this problem solved.

				-harry



On Mon, 20 Sep 2004, Jon Hanna wrote:

> 
> > Confusion does tend to creep in (as rdf stands now) when 
> > http: schemes 
> > are used in URIs used as identifiers.  Some people think that 
> > because an 
> > http: scheme is specific about how it is to be dereferenced, 
> > that when 
> > used as an rdf identifier it must be referring to the dereferenced 
> > representation.  But rdf actually says no such thing.
> 
> Certainly RDF says no such thing, arguably HTTP says no such thing
> either.
> 
> > IMHO, we just need a few predicates, and not another uri scheme.
> 
> Hopefully this is the last time I'll mention this, because I'm aware of
> quite a few flaws in it, but I made a stab at this idea at
> http://www.hackcraft.net/rep/rep.html if your browser doesn't do XSLT).
> 
> It's recently occurred to me that this can be made to do useful things
> with the "conceptual document" side of the HTTP-Range debate, and since
> the debate is the main reason I've put off doing anything more solid,
> and since I now think I can route around that, I'll hopefully find time
> to do so soon.
> 
> > Of course, I haven't yet read the "Web Proper Names: Naming 
> > Referents on 
> > the Web" paper
> 
> Please do. Either you'll agree with me and add another voice to my view
> of it, or you'll disagree with me and hopefully help me see the error of
> my ways (the authors of the paper have tried, but neither of us have
> budged much so far).
> 
> 
> 
>

Received on Monday, 20 September 2004 21:05:35 UTC