Re: Reminder - telecon to discuss test cases - tomorrow 11/20

Thanks,

in the discussion of dl-668 it was hypothesised that the test was wrong, 
having been derived from the faulty dl-208, more on that later.

Jeremy



Evren Sirin wrote:

> 
>  > Agenda for TEST subgroup meeting
> ...
>  > 6: Test with 1 pass
>  > overview and discussion
> ...
>  > - hard dl tests
>  > dl-666
>  > dl-668
> 
> I want to share the experience we had with our reasoner Pellet on these 
> test cases (dl-66x in general). Pellet cannot pass these test cases 
> right now. The ontologies in these test cases were originally OWL DL and 
> then a script converted the files to OWL Lite in a way that makes 
> reasoning really very hard for us (There are too many GCI's that we 
> cannot absorb). Since OWL Lite does not have complementOf, the script 
> creates two restrictions to define complement relation. Knowing how the 
> script worked, I've written another script that does the reverse 
> transformation. For example, the following description
> 
> Class(a:C10 complete restriction(b:P.16 maxCardinality(0)))
> Class(a:C10.comp complete restriction(b:P.16 minCardinality(1)))
> 
> is converted back to something like this
> 
> Class(a:C10)
> Class(a:C10.comp complementOf(a:c10))
> 
> After this transformation, these test cases become trivial for Pellet. 
> We still don't report success for these test cases but we know that we 
> can pass any dl-66x test after this transformation and believe they are 
> all ok. I thought this information would help the discussion about these 
> test cases.
> 
> Evren
> 
> ---
> 
>> Evren Sirin evren@cs.umd.edu
>> Graduate Research Assistant
>> Computer Science Department
>> Univ of Maryland, College Park, MD 20742
>> Phone: (301) 405-7027, Fax: (301) 405-6707
> 
> 
> 
> 
> 
>

Received on Friday, 21 November 2003 04:28:20 UTC