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

Re: results ontology

From: Jos De_Roo <jos.deroo@agfa.com>
Date: Thu, 11 Sep 2003 23:31:17 +0200
To: "Sandro Hawke <sandro" <sandro@w3.org>
Cc: www-webont-wg@w3.org
Message-ID: <OF222D588B.E57DC0AF-ONC1256D9E.0073A73A-C1256D9E.007647DA@agfa.be>

Sandro - indeed we shouldn't be able to pass
nonentailments and consistency tests and so
we took them out of our running and while at it
we also took out the running of importlevel and
notowlfeature tests as we think that what we
proved for those is not enough. Systems like
OWLP, Pellet, WonderWeb,... Jena? can do better.
We have updated our results

Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/

PS Jeremy, could it be that
   should have
   instead of

                      Sandro Hawke                                                                                                     
                      <sandro@w3.org>          To:       www-webont-wg@w3.org                                                          
                      Sent by:                 cc:                                                                                     
                      www-webont-wg-req        Subject:  Re: results ontology                                                          
                      2003-09-11 01:58                                                                                                 

Someone asked privately about Euler passing nonentailments and
consistency tests, which it shouldnt be able to do.

I'm curious about that, too.

I'm thinking of added a tres:note predicate, producing results like

      Pass [1]
      Pass [2]
      Pass [1]

where [1] and [2] would be links to explanatory or qualifying notes, where
people can explain why they think this is passing.   :-)

Also, I'm thinking of a bit of text at the top of each section
defining the term, so at the top of Consistency Tests it would say:
   "Pass" means returning "Consistent"
   "Fail" means returing "Inconsistent"
   "Incomplete" means returning "Unknown", not returning, raising
         and error, etc

... but as I write that I'm not even sure we'd agree (or, more to the
point, how OWL Test Cases reads on this).  Some people might consider
dumping core as a kind of Failure.

Of course we never say (I think) what kind of software might pass or
not pass an Entailment or Non-Entailment test, or what it might
return.   Ah well.

      -- sandro
Received on Thursday, 11 September 2003 17:35:08 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:56:55 UTC