- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Tue, 21 Jul 2009 12:44:29 +0100
- To: "Michael Schneider" <schneid@fzi.de>
- Cc: "Ian Horrocks" <ian.horrocks@comlab.ox.ac.uk>, "W3C OWL Working Group" <public-owl-wg@w3.org>
Just a POV. I've long found owl.owl (and rdf.rdf and rdfs.rdfs) to be positively harmful, when they've been attended to. This happens in three ways: 1) People do weird things with them (like import them); thankfully, this is nearly stamped out; they also trust them (wrongly) 2) Implementors use them in spite of their oddness/incompleteness (e.g., swi prolog, but perhaps this is a marginal implementation; an example where this sort of thing was, in fact, harmful, see the thread surrounding this message <http://lists.owldl.com/pipermail/ pellet-users/2007-August/001809.html> wrt the swrl ontologies). 3) It promulgates the idea that RDF or OWL is appropriate for this sort of thing. Which leads to more 1 and 2. It also fits in with some of my objections to GRDDL (and yes, I am well aware that some of these objection apply to the XML Schema...I'm not thrilled with the situation :)). I find the "linked data" rationale to be exceedingly weak, since this isn't data. Similarly, that some people like this style doesn't really do much for me since I suspect 1) they are in a very small minority and 2) it actually misleads them. That all being said, I won't object to this in spite of the irrationality of it. I accept we must kowtow to this sort of thing. I would suggest, however, that the file be *minimal*. Ideally, I would like *just* a list of terms with a seeAlso for each term pointing to the relevant part of the structural specification. I guess I'll go with "whatever", but I don't think it will represent the technical consensus of the group. Oh well. Cheers, Bijan.
Received on Tuesday, 21 July 2009 11:40:14 UTC