RE: Proposal: new top level domain '.urn' alleviates all needforurn: URIs

> -----Original Message-----
> From: ext Michael Mealling [mailto:michael@neonym.net]
> Sent: 10 July, 2003 18:38
> To: Stickler Patrick (NMP/Tampere)
> Cc: tbray@textuality.com; sandro@w3.org; hardie@qualcomm.com; 
> uri@w3.org
> Subject: RE: Proposal: new top level domain '.urn' alleviates all
> needforurn: URIs
> 
> 
> On Thu, 2003-07-10 at 11:07, Patrick.Stickler@nokia.com wrote:
> > > -----Original Message-----
> > > From: ext Michael Mealling [mailto:michael@neonym.net]
> > > On Thu, 2003-07-10 at 09:20, Patrick.Stickler@nokia.com wrote:
> > > > The purpose of the Semantic Web, and tools such as
> > > > OWL, is precisely to provide a standardized means of
> > > > providing such information to such applications.
> > > 
> > > Great to hear it! But how is that universally applicable 
> > > unless you are 
> > > suggesting that every application that uses URIs must become 
> > > OWL and SW
> > > compliant? I hope not. I don't think Cisco hubs can spare the 
> > > cycles for
> > > parsing RDF graphs.
> > 
> > I'm not proposing any such thing, only that RFC 2396 not
> > preclude the possibility that two lexically distinct URIs
> > might denote the same resource.
> 
> What definition of 'resource' are you using in that last 
> statement? The
> one found in 2396 or the one used by OWL/RDF/SW. They're two different
> things (and should be two different terms). Let's do that for 
> clarity's
> sake: A URI-Resource is the abstract thing that a URI is bound to. An
> SW-resource is a first order object that can make statements about
> itself such as its uniqueness, how its compared to others, etc. An
> SW-resource can also be a URI-Resource but when it is referred to as a
> URI-Resource it looses all of its SW-ish properties.

Hmmm... I'm not sure I completely follow you. How does a resource
make statements about itself?

I don't see that the definition of a resource needs to be
different for the Web versus the SW.

A resource is simply some entity that is referred to 
(denoted by) a URI.

Different layers are able to provide different functionality
relating to those resources in terms of URIs.

The Web provides facilities for obtaining representations
of those resources, and for indirectly linking one
representation to another by referring to resources
within representations. The link is not between the
resources, but between the representations.

RDF/OWL provide facilities for formally describing those 
resources.

The SW, based on standardized extensions to the Web such
as URIQA, provides facilities for obtaining descriptions
of those resources, and for indirectly linking one
description to another by referring to resources within
description statements.

Thus, for both the Web and SW, URIs denote resources.
That is the key point of intersection between the Web
and SW. How Web vs. SW machinery allow one to interact
with those resources, either by representations or
descriptions, does not require that the definition of
what a resource is and the function of a URI to denote
that resource must be different.

> I would then rewrite your last sentence like this:
> 
> "I'm not proposing any such thing, only that RFC 2396 not
> preclude the possibility that two lexically distinct URIs
> might denote the same SW-resource."

No. While that is also true. It is needlessly vague.

Just because RFC 2396 does not provide any means to determine
if two lexally distinct URIs do or do not denote the same
resource, does not mean that one must posit some different
kind of resource in order to allow SW machinery such as OWL
to do so.

> Your first statement is vague. My slight-rewording of it isn't 

Funny, I found your rewording far more vague than my version.

Though, the vagueness probably arises in the need to nail
down the definition of 'resource' (and ideally, IMO, a common
definition used by both the Web and SW).

I was glad to see Pat Hayes bringing this issue to the Tag
recently.

> and is
> now correct and does not attempt to make equivalence statements about
> URI-Resources when there is no method for doing so at that layer.

Here's where the miscommunication is, I think.

I was not asserting that RFC 2396 make any equivalence statements.

But that it would also not make any non-equivalence statements. I.e.
that it would remain completely silent about whether two lexically
distinct URIs *might* denote the same entity in the universe, 
whether or not one might get the same representations of that
same entity from both URIs via HTTP GET.

It's about RFC 2396 only saying what it must say, and not
hobbling the upper layers.

> > You seem to be asserting that they can't, at the level of
> > URIs alone, which hamstrings any higher layers.
> > 
> > > > Surely RFC 2396 is not intended to prohibit the
> > > > co-denotation of URIs?
> > > 
> > > Its not intended to prohibit it. But its not intended to 
> mandate it
> > > either. 
> > 
> > Well, all fine and good. But your earlier comments strongly
> > suggested otherwise.
> 
> Again, whenever you use the term 'resource' in an unqualified way I am
> assuming you are limiting your language to what is found in RFC 2396
> which has no language for stating equivalences for Resources. 

Well, that's hardly going to get us anywhere -- if I must constrain
my use of 'resource' to the meaning defined in RFC 2396 when the
very discussion relates to the meaning of 'resource' as defined
in RFC 2396. 

> If you are
> using 'resource' in another way then be clear about it.

I was trying to be. I hope my comments above make it clearer.
If not, I'll try again.

Patrick

Received on Wednesday, 16 July 2003 04:35:37 UTC