W3C home > Mailing lists > Public > www-rdf-interest@w3.org > April 2003

RE: Denotation of URIs

From: <Patrick.Stickler@nokia.com>
Date: Wed, 9 Apr 2003 09:33:31 +0300
Message-ID: <A03E60B17132A84F9B4BB5EEDE57957B5FBB7A@trebe006.europe.nokia.com>
To: <pfps@research.bell-labs.com>
Cc: <zednenem@psualum.com>, <www-rdf-interest@w3.org>

> -----Original Message-----
> From: ext Peter F. Patel-Schneider 
> [mailto:pfps@research.bell-labs.com]
> Sent: 08 April, 2003 20:54
> To: Stickler Patrick (NMP/Tampere)
> Cc: zednenem@psualum.com; www-rdf-interest@w3.org
> Subject: Re: Denotation of URIs
> From: <Patrick.Stickler@nokia.com>
> Subject: RE: Denotation of URIs
> Date: Fri, 4 Apr 2003 15:29:36 +0300
> > > -----Original Message-----
> > > From: ext Peter F. Patel-Schneider 
> > > [mailto:pfps@research.bell-labs.com]
> > > Sent: 04 April, 2003 15:00
> > > To: zednenem@psualum.com
> > > Cc: www-rdf-interest@w3.org
> > > Subject: Re: Denotation of URIs
> > > 
> > > 
> > > 
> > > From: David Menendez <zednenem@psualum.com>
> > > Subject: Denotation of URIs (was: URI for language identifiers)
> > > Date: Fri, 4 Apr 2003 02:13:02 -0500
> > > 
> > > > 
> > > > At 7:49 PM -0500 2003-04-02, Peter F. Patel-Schneider wrote:
> > > > >However, even
> > > > >in this very expansive notion of property there are 
> > > (still) considerable
> > > > >``fair use'' provisions.  Hopefully these provisions 
> will not be so
> > > > >weakened that I will be prohibited from making the 
> claim that the
> > > > >denotation of http://www.whitehouse.gov/#43 is Tipper 
> > > Gore's husband.
> > > > 
> > > > The only reason anyone might object to that would be if 
> there was 
> > > > already a widespread understanding that 
> > > > <http://www.whitehouse.gov/#43> denotes "The 43rd 
> President of the 
> > > > U.S."
> > > 
> > > But even so, one might want to argue that "The 43rd President 
> > > of the U.S."
> > > is not George W. Bush.  Requiring a common, fixed denotation 
> > > for all URI
> > > references would eliminate this possibility.  (Unless, of 
> > > course, you meant
> > > that there was some sort of intensional denotation here.)
> > 
> > Not at all. As I've given examples before, if you want to talk
> > about the denotation of a URI, you have to reify that URI and
> > make statements about the URI, not the resource denoted by that
> > URI.
> > 
> > You can't say 
> > 
> > <http://www.whitehouse.gov/#43> denotes "The 43rd President 
> of the U.S."
> > 
> > to mean that the URI http://www.whitehouse.gov/#43 denotes the 43rd
> > president of the US. What you've said above is that the resource
> > denoted by the URI http://www.whitehouse.gov/#43 denotes the 43rd
> > president of the US, or in other words, the 43rd president of the US
> > denotes the 43 president of the US.
> No, of course, you can't use URIs this way.  But I don't want 
> to, and don't
> have to.  Instead I could say
>    <http://www.whitehouse.gov/#43> owl:sameAs <name:AlGore> .
> which doesn't require reification or denotation relationships in the
> language.

Well, fair enough, but that's not actually what you said in your
example... so my comments are still valid re the example you used ;-)

> [...]
> > > I completely disagree with this.  If you nail down the 
> denotations of
> > > everything, then why bother with the Semantic Web?  
> > 
> > Eh? Huh? URIs are atomic primitives of the semantic web which allow
> > us to describe the *things* denoted by those URIs. The 
> point of nailing
> > down consistent unambiguous denotations is so that we can all talk 
> > about the same *things* with consistency and clarity (or at 
> least more
> > consistency and clarity than natural language).
> But people don't need to talk about the same things to be able to
> communicate.  There does need to be some shared context, but 
> this certainly
> does not need to be a total context shared by all.

It is my understanding that that is a commonly expected goal of
the SW. To provide just such a global shared context for communication,
and one that is much more precise than natural language.

> > I really wonder what kind of semantic web you are envisioning...
> One that is more liberal than the one you appear to be envisioning.


Though I wonder whether what you envision is also envisioned by
the majority of folks working on/with SW technologies. I don't 
think so.

> > You seem rather to be wanting a semantic web to define the 
> semantic web.
> > You appear to want a framework below the semantic web for 
> defining the atomic
> > primitives used by the semantic web. Well, fine and good. Go for it.
> > But until you provide it, the semantic web will have to operate on
> > the presumption that URIs are fully opaque, globally unambiguous
> > names for things and that a given URI always is *supposed* to mean
> > the same thing (not that there is any guaruntee is actually will).
> I'm not asking for this at all.  I even think that the 
> currently-proposed
> nature of the Semantic Web, as embodied in RDF and OWL, is 
> capable of doing
> much of this already.

Concrete examples are always welcome.

> > > It 
> > > appears to me that
> > > the benefits of the Semantic Web (over XML, for example) are 
> > > intimately
> > > involved with partial and conflicting information.
> > 
> > Sorry, but I will have to see a clear and motivating example of
> > such a benefit. What you are suggesting sounds irrational to me.
> >  
> > Show me the code... Show me a use case... Show me...
> Well most current Semantic Web code only works where the 
> context is shared
> in the way that you want.  I, however, am interested in 
> building Semantic
> Web applications that break out of this limitation.

Great. More power to you. But I think that most folks are not
interested in such applications, and certainly are not wanting
that in order for your applications to break out of such limitations
that their applications will simply break, by becoming unreliable
due to ambiguity.

See my final comments below.

> > And note specifically (and I've tried to stress this before) that
> > I differentiate between the denotation of a URI and knowledge
> > expressed about the denoted resource using that URI.
> I don't see a distinction here.

That may be why we fail to see eye to eye.

I consider the denotation of a URI to be atomic to the SW, and
outside the scope of the SW machinery, yet statements made about 
the resource denoted by a URI to be fully within the scope of
the SW and the primary focus/heart of the SW.

You'll have to provide some very persuasive examples/use cases/etc.
to convince me that there is no such distinction.

> For example, suppose that we have two web pages http://a.ex/ 
> containing
> 	<http://a.ex/john> rdf:type _:a .
> 	_:a owl:onProperty <http://c.ex/people#married> .
> 	_:a owl:maxCardinality "0"^^xsd:nonNegativeInteger .
> and http://b.ex/ containing
> 	<http://a.ex/john> <http://c.ex/people#married> 
<http://b.ex/mary> .

> What then is the denotation of <http://b.ex/mary>?  

I don't know. The denotation of URIs is outside the scope of SW
machinery. It denotes whatever the creator and/or owner of the URI says
it denotes.

> In particular, how is
> http://a.ex/ supposed to fit this denotation into its view of the world.

Sorry, but I don't see how this example relates to the SW if those
statements are merely the content of web pages.

Are those statements syndicated into a single RDF graph?

How does a web document have a view of the world?

Sorry. No comprende.

> > I agree that there can and will be disagreement between statements
> > about resources denoted by URIs, but there should not be disagreement
> > about the denotation of the URIs themselves. If there is, then 
> > how do you even know that you are disagreeing about statements
> > about a resource, if in fact, you are simply talking about
> > different resources?

> Maybe you won't even know that you are talking about ``different
> resources'' and you won't care.  

Oh, I think I will *always* care if there is such ambiguity.

> For example, you might disagree about who
> is the current first lady without disagreeing where the first lady lives.

Well, but if the URI being used to denote who the first lady
is, is being used ambiguously, then there is no way to actually
test for any disagreement. I.e., I may assert that _:a denotes
Jane and she lives at address X, but that _:b denotes Betty and
she lives at address Y. If you use _:a to denote Betty and assert
that Betty lives at address X, I will see no contradiction, because
according to my understanding of _:a you are saying that Jane lives
at address X!


Put that in my epitaph (eventually ;-)

> > Ambiguity in the SW is anathema, and will always be detrimental.

Yeah, what I just said. ;-)

> > Yes, SW agents will have to deal with it more or less, but the
> > less the better.
> >
> > It is *never* going to be beneficial.
> >
> > Though feel free to prove me wrong.

> Ambiguity is difficult to deal with, agreed, and it is reasonable to try to
> reduce it to the extent possible.  However, mandating that certain kinds of
> ambiguity are not permissable, just means that one cannot deal with
> situations where these kinds of ambiguity actually exist.  

I disagree. One can mandate all kinds of constraints yet still provide
machinery for dealing with violations of those constraints. A case in
point are HTML browsers. The HTML specs are very clear about what is or
is not a valid HTML instance, yet browsers are quite capable of dealing
with errors in HTML instances.

The same is true for SW agents. Ambiguity in denotation should be clearly
and strongly identified as bad, wrong, detrimental, etc. yet the SW 
architecture can still have machinery to help identify any occurring
ambiguity and deal with it productively.

Being able to deal with an error does not validate the error and make it OK.

Ambiguity in denotation will always be a bad thing. Always. Even if SW agents
can deal with it trivially (which I doubt they will be able to do short of
cognitive abilities).


Patrick Stickler, Nokia/Finland, (+358 40) 801 9690, patrick.stickler@nokia.com
Received on Wednesday, 9 April 2003 02:33:55 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:07:45 UTC