RE: reference needed

The same good'ol problem, but it seems to me more and more clearly set (or
maybe I just begin to understand it)

Zooming in on the problem and seeing the details is aproaching a solution -
hopefully.

Sandro does not buy Published Subjects as bringing anything into RDF realm,
and I concede they were not intended for that. Although Steve Pepper's
proposal for "Curing the Web's Identity Crisis" makes a step towards a
solution, it seems that this attempt is not meeting consensus of RDF folks,
if I understand well what I read at
http://esw.w3.org/topic/SubjectIndicator.

So, let's sum up some consensual things

1. URIs are identifiers - that's what the acronym expansion says, at least.
2. URIs can be used by different protocols (in a very broad sense of the
word "protocol", see below), among which http, to address and retrieve
"things" (chunks of bits, whatever)
3. URIs are considered as names in RDF and in OWL

But names are used for both identification and addressing , and Dan has a
very pertinent remark about it in
http://esw.w3.org/topic/WhenBrowsableAndUnambiguousCollide : it's accessing
always the same thing by a name that makes, by usage and experience, that
name an identifier of the thing. Based on trust, somehow.

It figures that there is no *absolute* identifier outside a defined context
of identification. I would say "protocol" if the word was not so much
loaded here. IOW, "identification" = "identifier + protocol", or
"identifier + context". And as Dan's remark shows well, for a name to be an
identifier is generally not a generic property, but the result of usage in
a certain protocol. Identification is a process of emergence. Difficult
stuff.

What do those considerations lead us to? That URIs, willy-nilly, have been
considered by usage as not-that-bad identifiers for things they were not
intended to identify to begin with, in emerging protocols. So they are de
facto overloaded, and identify differents things already. Like it or not.

Since we're bound to live with that, instead of fighting forever to know
what the hell a URI *should* identify, let's try to understand which
protocols of identification use URIs, and what *is* identified in each of
those protocols. Afterwards, it's syntactic stuff to know how the URI
should be dashed, hashed, mashed, prefixed or whatever to make the
different protocols somehow distinct.

For example, and to come back to PSIs, using a URI as "Subject Identifier"
explicitly uses a protocol where what is addressed through the URI is a
Resource (the Subject Indicator) indicating a Subject (a thing outside the
system). The identification protocol in that case is crystal clear.

In the context of http protocol, the URI identifies the Subject Indicator
(the information resource retrieved).
In the context of PSI protocol, the URI identifies the Subject itself.

Remark that a Subject Identifier is a URI used as a name of a thing exactly
like it is used in OWL. In my view, every owl:Class element could indeed be
a conformant Subject Identifier in PSI protocol. But PSI protocol demands,
to be effective, that the name has also addressing functionnality.

Does that help?

Bernard

Bernard Vatant
Senior Consultant
Knowledge Engineering
Mondeca - www.mondeca.com
bernard.vatant@mondeca.com


> -----Message d'origine-----
> De : www-rdf-logic-request@w3.org
> [mailto:www-rdf-logic-request@w3.org]De la part de Lynn Andrea Stein
> Envoye : mercredi 4 juin 2003 23:16
> A : Sandro Hawke
> Cc : www-rdf-logic
> Objet : Re: reference needed
>
>
>
> > Resent-From: www-rdf-logic@w3.org
> > From: Sandro Hawke <sandro@w3.org>
> > Date: Wed Jun 4, 2003  3:12:53  PM US/Eastern
> > To: www-rdf-logic <www-rdf-logic@w3.org>
> > Subject: Re: reference needed
> >
> >> but what if we're using
> >> an inferencing engine to reason about RDF statements? Would the engine
> >> treat all occurrences of SlashURIs as referring to the thing itself,
> >
> > All URIs in RDF refer to things themselves.  Those things themselves
> > might be web pages.
>
>
> But the URI (e.g., http://www.w3.org/Consortium/) construed in RDF (or
> RDFS or OWL)  can't simultaneously refer to
>
> 1) the bits returned by http get on that particular page
> AND
> 2) whatever happens to be the current description of the W3C
>
> let alone
>
> 3) the Consortium itself
>
> etc.
>
> In particular, if the bits returned are different tomorrow,
> interpretation (1) says either the URI still refers to the old bits OR
> the reference relation has changed -- the URI maps onto a different
> thing tomorrow -- while (2) says that the reference relation has
> remained the same while the referenced object has (internally) changed
> (its representation).
>
> So, while I agree completely that
>
> >
> > I think the RDF Model Theory is very clear that URIs (aka URIRefs)
> > function in RDF just like constant symbols in classical logic.  No
> > dereferencing is involved in knowing that each URI acts (within an
> > interpretation) as a name for something in the domain of discourse.
> >
>
> I've never been quite clear on which the (some)thing(in the domain of
> discourse) is that the URIRef names.  I suppose that I can use it
> however I want, but only at the risk of diluting the U -- universality
> -- in the URI.  And of course all three of (1) the bits returned (2)
> the (changing) current description and (3) the Consortium are things
> and so properly nameable by URIs....the question is just *which* URI
> (or *which* thing).
>
>

Received on Wednesday, 4 June 2003 19:07:57 UTC