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

Re: Terminology Question concerning Web Architecture and Linked Data

From: Pat Hayes <phayes@ihmc.us>
Date: Tue, 14 Aug 2007 11:56:30 -0500
Message-Id: <p06230903c2e78cb56117@[10.100.0.67]>
To: "Mark Baker" <distobj@acm.org>
Cc: "Stuart Williams" <skw@hp.com>, www-tag@w3.org

>On 8/13/07, Pat Hayes <phayes@ihmc.us> wrote:
>>  >Suppose Nadia had two email addresses, nadia-work@example.com and
>>  >nadia-personal@example.com.  She might use each to *directly*
>>  >identify, respectively, her work and personal email inboxes.  However
>>  >the organizers of a conference might be free to use either to
>>  >*indirectly* identify her person.
>>
>>  They are free to use either to refer to her. On the face of it, there
>>  in nothing to distinguish this kind of reference from any other kind
>>  of reference, seems to me. What is it, in your view, that makes this
>>  kind of reference "indirect"?
>
>It's that the conference organizers have no control over, nor insight
>into, exactly what the URI identifies.

But they do. They can presumably say that the URI indirectly refers 
to whatever they want it to. You yourself (below) assert that 
indirect reference is a form of reference: so, they get to say what 
the URI refers to.

>  So any statement made about
>what it identifies is guaranteed to be wrong in some contexts.

I have no idea what you mean by 'context' here. Who introduced 
contexts into a discussion of reference?

>
>>  >  If owl:sameAs were defined in terms of indirect
>>  >reference, then you'd be able to make silly declarations such as;
>>  >
>>  >   <http://www.number-10.gov.uk/> owl:sameAs
>>  >      <http://www.homeoffice.gov.uk/> .
>>  >
>>  >because both of those URIs can be used to indirectly identify the UK
>>  >government.
>>
>>  No, look, you can't have it both ways. Either 'indirect reference' is
>>  reference or it isn't.
>
>It is.
>
>>  If it is, then this assertion can be true, for
>>  the reason you give.
>
>It can be true in some contexts, but cannot objectively be true
>because an objective observer would see that the membership functions
>(from the REST definition of "resource") are different as a result of
>the representations returned from each URI being different in
>consistent ways over time.

None of this has the slightest relevance to OWL semantics, which do 
not mention contexts or 'objective' truth, or time. OWL is not a 
language whose meaning can be reinvented to suit each particular use: 
its semantics are defined mathematically and are normative, so if you 
use OWL syntax with a different semantics then you are not following 
the OWL specs. The OWL semantics says that AAA owl:sameAs BBB is true 
in an interpretation I when I(AAA)=I(BBB). That is a very 
straightforward, unambiguous definition. Now, as I understand your 
assertion above that indirect reference is a form of reference, this 
means that there an OWL interpretation in which URIs denote their 
indirect referents are indeed satisfying interpretations. In these 
interpretations, therefore, when I(AAA) is the indirect referent of 
'AAA', this equation will be true and so the OWL sameAs assertion 
will be true.

>I interpret the definition of owl:sameAs to be objective because it
>doesn't say anything about context of use.  Am I mistaken?

I have no idea, as I have no idea what you mean by 'objective' or 
'context of use'. Neither of these notions have any bearing on OWL 
meaning. It sounds like you should be using a different language from 
OWL altogether.

Pat

>
>Mark.
>--
>Mark Baker.  Ottawa, Ontario, CANADA.         http://www.markbaker.ca
>Coactus; Web-inspired integration strategies  http://www.coactus.com


-- 
---------------------------------------------------------------------
IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Tuesday, 14 August 2007 18:56:43 GMT

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