Re: closing ISSUE-22 (special syntax for role rule)

On Mittwoch, 23. April 2008, Carsten Lutz wrote:
> On Wed, 23 Apr 2008, Rinke Hoekstra wrote:
> > On 23 apr 2008, at 18:30, Markus Krötzsch wrote:
> >> I agree, but I think it is not quite so simple. The main issue here
> >> might be
> >> that, while many rules can be expressed in OWL2, some of these encodings
> >> would violate the structural restrictions without need. I suggest we
> >> consider
> >> at least some special cases of rules here to waive that restriction, and
> >> make
> >> room for future rule interfaces on top of OWL2. I agree that we should
> >> not make a new rule language (if anything, one would take OWL2 rules to
> >> RIF, I guess).
> >
> > Hi Markus,
> >
> > I guess I agree with you on this point... it would be a shame to have the
> > structural restrictions get in the way of something that *can* be
> > expressed without changing the semantics. Aren't we then just speccing an
> > ill-matched syntax?
> >
> > How big / numerous do you expect the special cases to be?
>
> To me, this sounds like opening Pandora's box. I guess there are loads
> of special cases that one could allow. Where to start and where to
> stop?  Moreover, people already find the non-structural restrictions
> awkward and difficult to understand. Now we want non-structural
> restrictions with (potentially a lot of) exceptions to them? <shiver>.

I agree. We should just be aware that there rules are not really expressible 
in common applications now. So it is not strictly syntactic sugar. Anyway, we 
have just agreed to *postpone* the issue to some later WG.

I also assume that the offending structural restrictions can be dropped 
completely for most profiles where they are not required anyway, which still 
leaves a lot of options for modelling rules.

>
> I like the work of Markus on rules, but I am sceptical to start poking
> holes into our non-structural restrictions. Too ad-hoc.

Sure, I did not mean to suggest that either. If anything, one might have 
considered a special syntax for "guarded role inclusions" that avoids the 
need of an artificial self-role. But less syntax is always good, too.

Best regards,

Markus

>
> greetings,
>  		Carsten
>
> > -Rinke
> >
> >> For people interested in a formal spec of a larger class of "OWL2
> >> rules", I point to the following works of ours on the topic:
> >>
> >> http://korrekt.org/page/SROIQ_rules
> >> http://korrekt.org/page/ELP
> >>
> >> The main work here is to show that one can use rules (hence many other
> >> OWL 2
> >> features) with our tractable profiles without hurting the polynomial
> >> reasoning. Moreover, there is also the Protege plugin by Francis Gasse
> >> (see OWLED-Washington papers, joint work with Volker Haarslev and Uli
> >> Sattler) to
> >> actually work with such rules -- maybe more concrete proposals could
> >> also emerge from that experience?
> >>
> >> Regards,
> >>
> >> Markus
> >>
> >> On Freitag, 18. April 2008, Michael Schneider wrote:
> >>> Hi!
> >>>
> >>> This has been an interesting exercise for me at that time, and helped
> >>> me to
> >>> better understand the power of sub property chains. It is nice to see
> >>> that something like this can actually be expressed within OWL 2 DL. But
> >>> directly
> >>> supporting this as a feature in the OWL language itself would look
> >>> rather strange to me.
> >>>
> >>> So I concur: +1 for REJECTING this issue.
> >>>
> >>> Cheers,
> >>> Michael
> >>>
> >>>> -----Original Message-----
> >>>> From: public-owl-wg-request@w3.org
> >>>> [mailto:public-owl-wg-request@w3.org] On Behalf Of Peter F.
> >>>> Patel-Schneider
> >>>> Sent: Friday, April 18, 2008 1:43 PM
> >>>> To: public-owl-wg@w3.org
> >>>> Subject: closing ISSUE-22 (special syntax for role rule)
> >>>>
> >>>>
> >>>> On 16 January 2008 Bijan added a note to the proposal for ISSUE-22:
> >>>>
> >>>> I think we should close this with no action. Here's why:
> >>>>
> >>>> 1) It's a new feature and there is no concrete proposal and I spent a
> >>>> few minutes trying to think of a syntax and had no good one other than
> >>>> the rule itself
> >>>>
> >>>> 2) Having just this one rule (which wouldn't be DL safe!) is very
> >>>> strange and might conflict with rule extensions
> >>>>
> >>>> 3) It seems that the best place for this is in a "Decidable swrl
> >>>> compiler" (as a visitor here was working on). There are *lots* of
> >>>> rules that you can compile using the new expressive property
> >>>> axioms. Why *this* one? Just because we thought of it? Better to
> >>>> encourage the development of these SWRL compilers and leave it to a
> >>>> "decidable fragments of SWRL" group.
> >>>>
> >>>> [Bijan Parsia]
> >>>>
> >>>> There does not appear to have been any futher discussion.
> >>>>
> >>>> I agree with Bijan's comments, and propose that ISSUE-22 be closed in
> >>>> this fashion.
> >>>>
> >>>> Peter F. Patel-Schneider
> >>>> Bell Labs Research
> >>
> >> --
> >> Markus Krötzsch
> >> Institut AIFB, Universität Karlsruhe (TH), 76128 Karlsruhe
> >> phone +49 (0)721 608 7362          fax +49 (0)721 608 5998
> >> mak@aifb.uni-karlsruhe.de          www  http://korrekt.org
> >
> > -----------------------------------------------
> > Drs. Rinke Hoekstra
> >
> > Email: hoekstra@uva.nl    Skype:  rinkehoekstra
> > Phone: +31-20-5253499     Fax:   +31-20-5253495
> > Web:   http://www.leibnizcenter.org/users/rinke
> >
> > Leibniz Center for Law,          Faculty of Law
> > University of Amsterdam,            PO Box 1030
> > 1000 BA  Amsterdam,             The Netherlands
> > -----------------------------------------------
>
> --
> *      Carsten Lutz, Institut f"ur Theoretische Informatik, TU Dresden     
>  * *     Office phone:++49 351 46339171   mailto:lutz@tcs.inf.tu-dresden.de
>     *



-- 
Markus Krötzsch
Institut AIFB, Universität Karlsruhe (TH), 76128 Karlsruhe
phone +49 (0)721 608 7362          fax +49 (0)721 608 5998
mak@aifb.uni-karlsruhe.de          www  http://korrekt.org