W3C home > Mailing lists > Public > www-webont-wg@w3.org > May 2003

Re: comments on OWL test cases

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Thu, 15 May 2003 09:01:18 +0300
To: www-webont-wg@w3.org
Message-Id: <200305150901.18175.jjc@hpl.hp.com>

Ian if you have time please have another look at
http://www.w3.org/2002/03owlt/editors-draft/draft/#consistencyChecker

The changes made since the version last week are:
- new conformance statement requiring documentation of supported datatypes 
(stating the obvious)
- new conformance statement concerning correct result with Unsupported 
datatypes (Unknown in certain circumstances).
- s/complete/decisive/
- deletion of "otherwise it is [in]complete"
- (indentation typo) a new conformance clause SHOULD NOT return Unknown
  (again stating the obvious, mainly motivated to ensure that the royalty free 
nature of this WG covers non-trivial techniques)


On 'decisive'
> An OWL consistency checker is complete if, whenever its input document
> is [in]consistent, it would eventually (given unlimited, but finite,
> resources of memory and time) return the answer '[In]consistent'.

Pat points out that this use of the word complete is non-standard; (being less 
up on the terminology than Pat, I was surprised at this assertion, and found:

http://www.ai.sri.com/snark/tutorial/node3.html
[[
In this configuration SNARK is a logically complete theorem prover; that is, 
if conclusion does follow from the assertions, SNARK will find a proof. 
Therefore, we know that in this case the assertions do not imply the 
conclusion. 

On the other hand, if SNARK runs on for longer than we expect, there is no 
way, in general, to determine if the conclusion is not valid or if we simply 
haven't given it enough time.
]]

We are trying to devise a term for what we have been calling a 
complete OWL Lite Consistency Checker
and (unless Dan gets his way)
complete OWL DL Consistency Checker

but what we mean is not what is written in the SNARK documentation (which I 
believe is the normal usage).
Since SNARK is a potential base on which an OWL consistency checker might get 
written, and noting that Sandro was confused about our terminology here (I 
think because it was non-standard), I am fairly convinced that Pat has a 
point.

I am happy to support proposed minor edits (such as %s/decisive/foobar/g ) at 
this point, but if we need a rewrite of this section then we will have to 
postpone the last call vote.

Jeremy
Received on Thursday, 15 May 2003 04:41:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:00 GMT