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

Re: Sketch: reasoning conformance levels

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Date: Thu, 07 Nov 2002 11:10:08 -0500 (EST)
Message-Id: <20021107.111008.53877511.pfps@research.bell-labs.com>
To: jjc@hplb.hpl.hp.com
Cc: www-webont-wg@w3.org

From: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>
Subject: Sketch: reasoning conformance levels (was RE: Issue: Add hasValue to OWL Lite)
Date: Thu, 7 Nov 2002 15:35:01 +0100

[...]

> Reasoning components MAY claim "OWL DL reasoning" (aka "complete OWL DL
> conformance") if they provide complete reasoning over OWL DL. i.e. An "OWL
> DL reasoner" MUST find proofs for all OWL DL inferences. An OWL DL reasoner
> MAY find proofs for any OWL full deduction.

This has to be modified to handle imports constructs.  I suggest that it
say something like 

  Reasoning components MAY claim "OWL/DL entailment" (aka "complete OWL/DL
  conformance") if they compute all entailments between OWL/DL documents
  that do not contain imports.  Any reasoning component that claims "OWL/DL
  entailment" must somehow signal that its reasoning may be incomplete if
  it cannot find an imported document.  An OWL/DL reasoner MAY find proofs
  for any OWL/Full entailment.

I prefer using a / so as to prevent confusion in constructs like ``OWL
Full entailment''.

Also, there is no definition for inference or deduction in OWL, only for
entailment.  (I just went through and removed the single mention to
inference or deduction in the semantics document, not to confound Jeremy,
but because I just realized that I should not mention inference or
deduction in the semantics document.)

> Reasoning components MAY claim "OWL Lite reasoning" if they provide OWL Lite
> conformance (i.e. no OWL Lite constructs makes the reasoner fall over, and
> name separation is supported) and the reasoner will find proofs for at least
> ... [tbd]. An OWL Lite reasoner MAY find proofs for any OWL Full deduction.
> 
> Reasoning components MAY claim "most of OWL DL reasoning" if they provide at
> least OWL Lite reasoning and ... [tbd] (e.g. pass 90% of the tests).

This is, in my opinion, a *terrible* idea.   Reasoning components may, if
they wish, describe what sort of reasoning they do, but passing 90% of an
arbitrary and changeable set of tests is not a useful description, and
certainly should not be sanctioned as an official way of describing a
reasoner.

[...]

> Jeremy

peter
Received on Thursday, 7 November 2002 11:10:20 GMT

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