empty Thing (was Re: proposal to approve tests)

> > Thing-001       (unless Ian wants us to hold off still)

> I do. I would like to change the semantics of OWL DL/Lite so that it
> requires the domain to be non-empty. According to my reading of the
> OWL Full and RDF semantics, this should already be the case for OWL
> Full, and so as well as being more consistent with standard DL and FO
> semantics, it would also fix a mis-match between DL/Lite and Full
> semantics.


OWL Full requires Thing to be infinite ...

if in contrast you want Thing to be at least n-elements big (with perhaps n = 
1) then

EnumeratedClass( owl:Thing, uri-1, uri-2, .... uri-n )

is a small test case that exercises this difference between DL and Full.
(consistent in DL, inconsistent in Full)

The current situation, with n=0, permits small examples that exhibit this 
difference. Also in OWL DL 0 is already a special number, 1 less so.

Summary,
   if you propose an infinite Thing I would support it, despite the lateness 
of the change. I don't much like any finite number but zero is better than 
one, and also the change seems insufficiently motivated at this point - since 
the alignment with Full is specious.

Jeremy

Received on Wednesday, 29 October 2003 13:36:12 UTC