- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Mon, 14 Jan 2008 09:54:11 +0000
- To: Bijan Parsia <bparsia@cs.man.ac.uk>
- CC: OWL Working Group WG <public-owl-wg@w3.org>
Hmmm, interesting .... a very simple question. Bijan Parsia wrote: > > On Jan 12, 2008, at 11:52 PM, Alan Ruttenberg wrote: > >> Testing my understanding of the skolem constant proposal for anonymous >> individuals: >> >> If we adopted the treatment of bnodes as skolem constants, would it be >> the case that we no longer had the tree-shape restriction for related >> anonymous individuals? > > Correct. You could still get *them*, but you would have to encode them as The *them* I marked refers to what. I found I couldn't understand this sentence. But the rest of the message looked very promising. Maybe give an example, with a non-tree, and explain how it differs. > existential restrictions, or using Carsten's proposed universal role. > ==== I have no in principle opposition to skolemization. On the contrary, given that many implementations use the technique, the specs should, in my view, permit such an implementation (and do so explicitly if there is any room for doubt). I always understood the OWL 1.0 specs as permitting such an implementation. However, the relationship with RDF semantics, which treats bnodes as existentials rather than skolems, needs to be clear. [Concerning the HP position, I am not wholly sure what it would be, and if there was what may be thought of as a significant change, I would need to refer back to my colleagues. My expectation at this stage is that there may be differences of opinion, resulting in a formal abstention from HP. i.e. at this stage, I can't promise to deliver an HP vote, but Bijan's message seems to indicate a design change that addresses some of HP's concerns] Jeremy
Received on Monday, 14 January 2008 11:27:02 UTC