- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Thu, 9 Oct 2008 16:08:57 +0100
- To: Ivan Herman <ivan@w3.org>
- Cc: Boris Motik <boris.motik@comlab.ox.ac.uk>, 'Alan Ruttenberg' <alanruttenberg@gmail.com>, 'W3C OWL Working Group' <public-owl-wg@w3.org>
On 9 Oct 2008, at 10:53, Ivan Herman wrote: > Hi Bijan, > > I take your point on the practical usage of range/domain, thanks. If I > understand it well, at least this use case does not really care about > the possible OWL inferences, because separate tools would look at and > handle annotations anyway. They certainly don't care about the "integrity constraint" version. They usually don't *mind* e.g., getting a contradiction if they stick an integer into a string range, but that's not the primary job. (Indeed, one evil trick is to put a min 0 for some property to indicate that that class "can have" such a property. I just cannot convince Alan that this doesn't actually do what he thinks it does ;)) > Coming back to the original discussion, Boris showed in > > http://lists.w3.org/Archives/Public/public-owl-wg/2008Oct/0016.html > > that the simple approach (adding, SubAnnotationPropertyOf, > AnnotationPropertyDomain and AnnotationPropertyRange into the func. > spec > with an obvious mapping to the existing RDF vocabulary) might lead > to a > semantic divergence between OWL DL and OWL Full. What your use case > suggests, though, that users might not really care too much, because > they would not consider the OWL reasoning on these anyway... Ie, we > can > just live with that. Yep. Cheers, Bijan.
Received on Thursday, 9 October 2008 15:06:14 UTC