- From: Ian Horrocks <ian.horrocks@comlab.ox.ac.uk>
- Date: Fri, 3 Apr 2009 14:20:48 +0100
- To: Sandro Hawke <sandro@w3.org>
- Cc: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>, public-owl-wg@w3.org
On 2 Apr 2009, at 13:38, Sandro Hawke wrote: > >>> Remove EdNote on GRDDL. >> >> I did that, because AFAIK we did decide exactly what to do about >> GRDDL (see http://www.w3.org/2007/OWL/tracker/issues/97). It was >> subsequently determined that we can have multiple transforms, and >> also that Sandro will develop a suitable XSLT (with some technical >> help from Bijan). Others may have a different view? > > While that's all true, we did agree (although not "resolve") in a > telecon to add that note AFTER we had closed the issue and settled on > our current course of action. I think it was Bijan who was arguing > for > this ednote, though, so if he doesn't have a problem with removing it, > then... okay. Seems that Bijan doesn't have a problem (well, not this problem anyway). > >>> 2.4/ Profiles: >>> >>> Remove Figure 2, as it serves no useful purpose. > > I'm tempted to bet folks a drink that we'll get complaints about > removing it, but whatever.... I'm baffled as to what purpose you think this figure serves and why its loss will be mourned. AFAICT, the information content of the figure is: * Full > DL > RL,QL,EL * RL, QL and EL have a non-empty intersection Both of these seem trivially obvious and/or are already stated in the document. W.r.t. the first point, the first paragraph of 2.4 already says "Each profile is defined as a syntactic restriction of the OWL 2 Structural Specification, i.e, as a subset of the structural elements that can be used in a conforming ontology, and each is more restrictive than OWL DL." W.r.t. the second point, it's really impossible to imagine a design where the intersection would be empty. Moreover, as was already pointed out, the figure might raise more questions than it answers, particularly w.r.t. the *exact* contents of this non-empty intersection. Finally, we never even referred to this figure in the text, which I also take as a strong indication that it could be removed without loss of information. > >>> 3/ Differences >>> >>> Could remove the subsection headers, as the subsections are all very >>> short. >>> >>> >>> Change "albeit under a possibly different name." to >>> "albeit possibly under different names." >>> >>> Remove "; it also has a formal equivalence to UML [UML]." >>> This is just *wrong*. >>> >>> 3.2: >>> Just put this stuff elsewhere (perhaps in Primer). >> >> I significantly shortened the whole of Section 3 and pointed to NF&R >> for detailed explanation/documentation. >> >> I also renamed it "Relationship to OWL 1" as this seems more >> appropriate and avoids the negative connotations of "differences". > > Very nice, except that we need a link explaining the "almost"s in > paragraph two to a place with text like Peter and I were crafting > yesterday. If I were an OWL 2 user, I would insist the text > actually be > normative, too. (I guess there's no problem with a little normative > text in NF&R.) > > Am I the only one who thinks OWL 1 users will want to know, in no > uncertain terms, whether OWL 2 breaks their stuff, BEFORE they accept > OWL 2? This is *exactly* what I am trying to achieve here. > Figuring that out by sifting through our entire spec seems a bit > much to ask. That isn't the intention. The intention is to tell them that everything is fine, *which it is*. IMHO this message, we should *not* provide irrelevant information about the corner cases and "bug-fixes" in OWL 2 that prevent us from simply saying that it is *completely* backwards compatible with OWL 1 -- this does need to be documented somewhere, but not here (not sure if it should be normative and/or in NF&R, but this is a different issue that I will address in another email). I changed what is said here to make the message even more clear: it now says that "backwards compatibility with OWL 1 is, to all intents and purposes, complete" and that inferences are identical "in all practical cases". I also added a note explaining that "even the theoretical possibility of different entailments arises only from a change in the treatment of annotations in the Direct Semantics [OWL 2 Direct Semantics] that reflects what was typically implemented in OWL 1 systems". I'm ambivalent about this note -- we could simply say "see XXX for more details". BTW, given that the differences in entailments only affect OWL DL and derive from changes in the Direct Semantics, it seems to me that this is the right place to document them. Ian > > -- Sandro
Received on Friday, 3 April 2009 13:21:44 UTC