W3C home > Mailing lists > Public > www-webont-wg@w3.org > November 2002

RE: MT for imports (was: Re: Imports Proposal)

From: Massimo Marchiori <massimo@w3.org>
Date: Wed, 13 Nov 2002 17:05:52 +0100
To: "pat hayes" <phayes@ai.uwf.edu>, "Jeff Heflin" <heflin@cse.lehigh.edu>
Cc: <www-webont-wg@w3.org>
Message-ID: <NGBBJNKIMLOPPCFHEJEMCEAHDBAA.massimo@w3.org>

Yes, that's precisely the major unanswered issue I was pointing out in
http://lists.w3.org/Archives/Public/www-webont-wg/2002Nov/0117.html
(and previous emails)
"owl entailment would depend on the timed web structure"
and raised as very first objection at the first telecon we had on this ;)

There's more to this issue than the simple tracking, that's what I meant at the end of
http://lists.w3.org/Archives/Public/www-webont-wg/2002Nov/0134.html .

However, before spelling them out properly, the best thing is just to apply those consequences to the last wording of the proposal,
which alas I didn't have the time to read.

Anyway, in a nutshell, putting this into the entailment relationship means that you have troubles unless you extend the basic RDF
model. As it is now, not having time information labels, you CAN'T deduce the
(A imports B) -> "A plus the things in B"
(informally speaking)
because the time variable present on the left is not present any more in the right (no time in RDF yet...). Which means this is
false, UNLESS you force ontologies to be immutable.
Which is a possible design choice, but I think it's an ugly choice.
That's why operational import would be preferred here: it takes the mess out of the logical entailment, and allows us to properly
define a cleaner model in a v2.

-M

> -----Original Message-----
> From: www-webont-wg-request@w3.org
> [mailto:www-webont-wg-request@w3.org]On Behalf Of pat hayes
> Sent: Wednesday, November 13, 2002 3:47 PM
> To: Jeff Heflin
> Cc: www-webont-wg@w3.org
> Subject: Re: MT for imports (was: Re: Imports Proposal)
>
>
>
> Jeff, your email got me thinking about the intricacies which arise
> when thinking about imports in the context of a changing world. Here
> are few more example scenarios. In each case A, B, C etc are people,
> P, Q, R, etc are chunks of OWL in documents. 'changes' means altering
> the RDF at a given URL.
>
> 1.
> A publishes P
> B publishes Q importing P
> A changes P (to P')
> C reads Q and imports P'
>
> Now, has C got it right, or not? Or should C have imported P (how?)
> Or should B have tracked A's changes (how?)
>
> 2.
> A publishes P
> B publishes Q
> C publishes R importing Q
> B changes Q to Q' importing P
> D reads R
>
> Has D got it right? This is really a special case of the first one,
> but since the change involves an imports, the effect is magnified, as
> it were. Obviously, the change could be arbitrarily far along an
> imports-reference chain.
>
> 3.
> A publishes P
> B publishes Q importing P
> A's server crashes
> C reads Q , concludes that the imports P is empty, archives the result
> A's server comes back online
>
> Now has C got it right? Or should C have refused to archive an
> empty-due-to-404  imports statement?
>
> Pat
>
> --
> ---------------------------------------------------------------------
> IHMC					(850)434 8903   home
> 40 South Alcaniz St.			(850)202 4416   office
> Pensacola               			(850)202 4440   fax
> FL 32501            				(850)291 0667    cell
> phayes@ai.uwf.edu	          http://www.coginst.uwf.edu/~phayes
> s.pam@ai.uwf.edu   for spam
>
Received on Wednesday, 13 November 2002 11:06:42 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:55 GMT