W3C home > Mailing lists > Public > www-rdf-logic@w3.org > June 2003

Re: reference needed (URIs and what they refer to)

From: Drew McDermott <drew.mcdermott@yale.edu>
Date: Fri, 6 Jun 2003 13:10:42 -0400 (EDT)
Message-Id: <200306061710.h56HAgS12002@pantheon-po03.its.yale.edu>
To: www-rdf-logic@w3.org

   From: "Bernard Vatant" <bernard.vatant@mondeca.com>
   Cc: "tm-pubsubj" <tm-pubsubj@lists.oasis-open.org>
   Date: Thu, 5 Jun 2003 09:38:26 +0200

   ... To complete my previous post, I would add that your
   original diagram at
   is exactly isomorph to the schema PSI specification propose. Just a
   question of matching vocabularies.

   In your diagram:

   URI ---> identifies ---> Resource
   Resource ---> is represented by ---> "bag of bytes + media type"

   Matches the following schema in PSI terminology. Remember "Subject" is the
   Topic Map term for RDF "Resource".

   Subject Identifier (URI) ---> identifies ---> Subject
   Subject ---> is indicated by ---> Subject Indicator

   I guess "indication" and "representation" are the same here. It is for
   human consumption, to figure what the Subject-Resource *is" ....

It's important to be clear about this, because the "representation" of
a proper name (which is what an URI is) is a rather fuzzy concept.
It's certainly not a definition of the name, and may or may not
actually suffice to identify a person.  For instance, the
representation of "Hendrik Lorentz" may be "the scientist who devised
the Lorentz transformations."  This phrase may mean almost nothing to
the person who reads it, but nonetheless may suffice to distingish him
from Konrad Lorenz or some other person.

Many philosophers would say that proper names work by "baptisms" plus 
causal transmission, mainly through social mechanisms.  That is, the
person in question was officially dubbed "Hendrik Lorentz" at one
point, was introduced as such throughout his lifetime, was so called
whenever someone wanted to talk about him, and so forth, until
everyone is now pretty sure that when they use the name they will be
taken to be referring to that guy.

The good thing about URIs is that they provide a useful syntax for
proper names.  It suggests, for instance, who is responsible for
introducing a name (the host part of the URI).  However, it doesn't
really advance our understanding of the semantics of proper names, and
introduces confusion when a URI happens to be a URL as well.  It would
be nice if there were a strict syntactic separation between the
two. For one thing, a URL is in some ways not a proper name at all,
but a precise description of how to find the object referred to.  That
is, http://www.foobar.org/index#buzz can be taken to refer to a
location in a web page reachable by using a particular protocol with a
particular web host.  If the object at that location changes, the
denotation of the URL changes.  No similar property holds for URIs,
because they don't refer to locations at all.

                                             -- Drew McDermott
Received on Friday, 6 June 2003 13:10:44 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 2 March 2016 11:10:40 UTC