W3C home > Mailing lists > Public > www-tag@w3.org > July 2007

RE: Terminology Question concerning Web Architecture and LinkedData

From: Conal Tuohy <conal.tuohy@vuw.ac.nz>
Date: Wed, 25 Jul 2007 11:23:43 +1200
To: "Booth, David (HP Software - Boston)" <dbooth@hp.com>
Cc: Mark Baker <distobj@acm.org>, Dan Brickley <danbri@danbri.org>, Chris Bizer <chris@bizer.de>, www-tag@w3.org, semantic-web@w3.org, Linking Open Data <linking-open-data@simile.mit.edu>
Message-Id: <1185319423.14697.161.camel@localhost>

On Tue, 2007-07-24 at 14:19 -0400, Booth, David (HP Software - Boston)
wrote:
> > From: Conal Tuohy
> >
> > On Tue, 2007-07-24 Booth, David (HP Software - Boston) wrote:
> > > . . .  For the
> > > most part, the only way one can be sure that two URIs really
> > > do name the same resource is if they are provably defined to
> > > do so, such as: (a) if they are the same URI;
> > > (b) if one is declared by its owner to be owl:sameAs the other; 
> > > or (c) if the URI declarations are exactly the same.
> >
> > Why (in b above) only "by its owner"?
> > Is it because only the owner of URI A "really knows" what URI A
> > means? 
> 
> It is because only the URI owner has the authority to declare what
> resource that URI names, i.e., to define the association between the URI
> and a particular resource.  See WebArch section 2.2.2:
> http://www.w3.org/TR/webarch/#uri-assignment

Are you saying it's purely a terminological matter? Isn't it also an
empirical question about correctly identifying individuals?

> This does *not* mean that the URI owner also owns the resource.  For
> example I could mint a URI that identifies your car.

You could ... though I don't have a car. What would your URI really
identify, I wonder? :-)

> > If so, how does the owner of URI A know what URI B
> > "really" means?
> 
> Hopefully, by dereferencing B to find an authoritative URI declaration
> for B, i.e., a declaration that defines the association between B and a
> particular resource.  If B cannot be dereferenced, then you might need
> to call B's owner on the phone and ask.

Even so, you could make an honest mistake in your understanding of my
URI B, and wrongly equate your URI A with a URI B which actually
identifies something else. 

You may make a number of correct assertions using your URI A on the
understanding that it refers to person A, but make a mistaken owl:sameAs
assertion equating it with URI B. Are you saying the owl:sameAs
assertion trumps the other assertions? i.e. you may have intended to
talk about person A, but by equating your URI A with URI B, you have
"actually" (by definition?) been talking about person B?

If you take the view that the URI owner is by definition correct
whenever they use owl:sameAs to identify the subject of their URI, what
happens if I mint a URI (e.g. to represent Tim Berners-Lee, who seems
fated to serve as a perennial example) and equate it with multiple
existing URIs for him, using owl:sameAs, but in the course of this
"authority work" I make a mistake and include one URI which actually
identifies some other person? 

While it's true that the web architecture says that the owner of a URI
has the right to define what it identifies, URI ownership is not a magic
shield against making factual errors in defining what the URI
identifies. 

C
Received on Tuesday, 24 July 2007 23:27:26 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:46 GMT