Re: TAG request: establish the relationship between URIs and Resources is many to many

Bill de hÓra <dehora@eircom.net> wrote in 
http://lists.w3.org/Archives/Public/www-tag/2003Jan/0313.html :

>Tim Bray wrote:
> > . . .
> > 2. The Web is designed on the principle that a single URI identifies a
> > single resource which does not change.  In practice, this principle is
> > someties violated (insert list of nasty examples), and software must
> > often deal with the consequences, but such inconsistency is always
> > damaging and SHOULD be avoided.  -Tim

It's true that I shouldn't try to use the same expression to refer to both
the car (in one statement) and the picture of the car (in another
statement).  But the point is
that it's very helpful (and "good practice" according to the TAG), that if
you're using a URI to identify something that does not exist directly on
the Web (such as a car), you should place a document at the end of that
URI, which describes the thing that you're using that URI to identify.   In
TimBL's example, that document is a picture of the car.  The car and the
picture are two different things.  Yet somehow I should be able to make
unambiguous statements about either one.  For example, I should be able to
make two statements of the form:

        X #is #beautiful
        Y #is #beautiful

(where X and Y represent some kind of designation involving the URI) and be
unambiguously referring to the car in the first statement, and the picture
of the car in the second statement.
Therefore I need a way to distinguish which one I mean when I make a
statement.  There are two ways that can be done -- different names or
different context -- and they have different pros and cons.

>Clearer, thanks. But abuse of this principle also affects
>specifications such as RDF, as well as web software.

The issue isn't abuse.  The issue is proper but potentially *different*
use by two parties that could cause ambiguity if they don't happen
to use the same conventions.

>For example it
>will mean we have to merge RDF graphs with great caution before
>inferences can be made and we have to be careful about RDF queries
>that span multiple graphs.
>
>If the rdf-wg were happy to add words to the primer about how
>breaking this principle interplays with the function that determines
>the denotation of a URI, that would help greatly. That is, when
>merging two RDF graphs, be aware that a URI  used in graph 1 might
>not have the same denotation as it does in graph 2.

Exactly.  But that is NOT acceptable for the Semantic Web, because it
would inhibit the "View Source" effect, which critically depends on
using the same or very similar URLs to indicate two different things.
This is explained more fully in
http://www.w3.org/2002/11/dbooth-names/dbooth-names_clean.htm#EnablingViewSource



-- 
David Booth
W3C Fellow / Hewlett-Packard
Telephone: +1.617.253.1273

Received on Monday, 27 January 2003 00:13:53 UTC