Re: Issue: Add hasValue to OWL Lite

I see the following options emerging from the discussion surrounding adding hasValue
to OWL Lite.
This attempts to choose highlights from the email discussion with the subject OWL Lite
semantics as well.

1 - do not add any notion of hasValue to OWL Lite.
        con:  after releasing the OWL lite description, the public comments supported
the notion of adding hasValue to OWL Lite since there are claims that with this one
feature added to OWL Lite some important use case applications could be represented in
just OWL Lite.
        pro:  does not impact our timeline and does not make OWL Lite any heavier.

2 - add hasValue to OWL Lite with the semantics as specified in OWL DL.
        con: makes OWL Lite heavier.  Some would claim too close to OWL DL in
computational complexity, difficulty to implement, and difficulty to explain to users.

        pro: addresses the public comment requests to add hasValue and introduces no
new semantics to explain.

3 - add hasValue to OWL Lite with a restricted semantics.  A restricted semantics was
proposed by Jeremy.
    con:  makes OWL Lite a little heavier but more importantly adds a new semantics to
explain which adds additional complexity to the documents and tutorials and may cause
confusion.
    pro: addresses the public comment requests for an addition of hasValue by adding a
limited notion of hasValue.  (Some claim the specific limited notion of hasValue does
not make the implementation task of adding hasValue as difficult but this claim has
been disputed.)

A personal observation of the comments so far:

While I applaud the work that has been done quickly on option 3, I think the strongest
argument against it so far may be paraphrased as the following:
    it has taken 9 months or more to obtain a semantics for OWL that we can agree on
and it is unlikely (you could read this as impossible) to generate and agree on a
semantics for option 3 in the time remaining.  Thus option 3 is infeasible given our
time frame.    I find this argument hard to dispute and thus find it difficult to
support option 3.

concerning option 2, the line of reasoning that says adding owl dl semantics hasValue
to OWL Lite makes it either too hard to implement or too close to OWL DL may mean that
some do not bother to make an effort to implement OWL Lite  or they do and they
generate incomplete OWL Lite implementations (thereby possibly making up their own
alternative semantics).
While that could be problematic, I was willing to live with it since I also believe
hasValue is critical for many applications and I believe some complete implementations
would be available (and probably some incomplete implementations would be generated as
well).
This allowed me to not explicitly bless some particular restricted semantics (as
option 3 does) but did mean that I realized that some limited implementations were
likely to arise.   This position could be seen as a hedge position or a complete cop
out (if someone is going to specify a restricted semantics then our group should be
the one to do it rather than leaving it up to developers presumably less knowledgeable
than we are in the area).
If it is the only path to having the group move forward, i will acknowledge that my
position on option 2 could be seen as the cop out position and return to option 1.

deborah

Ian Horrocks wrote:

> On December 9, Jim Hendler writes:
> >
> > At 8:09 PM +0000 12/9/02, Ian Horrocks wrote:
> > >This has been overtaken by Jeremy's proposal and the resulting
> > >discussion - see [1]
> > >
> > >Ian
> >
> > Ian - forgive my stupidity, but I'm not sure I see how Jeremy's email
> > and the resulting discussion answers the question that Deb is asking.
> > Are you claiming it breaks this thing about OWL entailments?
> >   If so, then we go back to Deb's original proposal to have hasValue
> > in Lite, but with IFF semantics.  I understand you still would oppose
> > that, but I'm just trying to make sure I understand - is the
> > following a correct sumary of your position?
> >   i. You oppose adding hasvalue with if in Lite, becuase it has iff in
> > DL (c.f. your response to Jeremy)
>
> I oppose giving Lite a different semantics from *both* the other two
> languages. I hope that the responses from Peter, Pat and myself make a
> clear argument as to why we should NOT consider doing this, in
> particular given the stage we are at in the overall process.
>
> >   ii. You oppose havine hasvalue with iff in Lite, because you argue
> > it pushes up the computational complexity of Lite
>
> I oppose adding hasValue to Lite because it would make it almost
> exactly equivalent to OWL DL, i.e., it would have the same
> computational complexity, and more or less the same implementational
> complexity. Adding hasValue would therefore be completely pointless -
> we should just scrap OWL Lite.
>
> Ian
>
> >    If I've gotten this wrong, please help -- I think it important we
> > understand this issue well, because I expect we will be voting on it
> > this Thursday (as I see no consensus emerging)
> >
> >   -JH
> >
> >
> > >
> > >[1] http://lists.w3.org/Archives/Public/www-webont-wg/2002Dec/0054.html
> > >
> > >
> > >On December 5, Deborah McGuinness writes:
> > >>
> > >>  >
> > >>
> > >>  In preparation for next week's discussion of hasValue, in going
> > >>through the mailing list, this is the only
> > >>  written compromise solution.  My reading of this is a proposal that says:
> > >>
> > >>  MexicanRestaurant = (hasValue foodServing MexicanFood)
> > >>
> > >>  then if x is known to be an instance of mexicanRestaurant then we
> > >>know that x has a value of MexicanFood
> > >>  for its foodServing slot
> > >>
> > >>  (i.e., inheritance works because hasValue is a necessary condition)
> > >>
> > >>  BUT
> > >>  if y is known to have MexicanFood for a value of its foodServing
> > >>slot then we can NOT infer that y is an
> > >>  instance of MexicanRestaurant.
> > >>  (i.e., recognition does not work because hasValue is not
> > >>considered to be a sufficiency condition).
> > >>
> > >>  this proposal means that hasValue has a different semantics than
> > >>it has in DAML+OIL and we would need to
> > >>  decide what the semantics would be in OWL DL and OWL Full.
> > >>
> > >>  I was not on a telecon where I believe this issue may have been
> > >>discussed more completely and if someone
> > >>  who was on that call can relay any consensus, that would be useful
> > >>in preparation for next week's
> > >>  discussion of this issue.
> > >>
> > >>
> > >>  > Re: Issue: Add hasValue to OWL Lite
> > >>  >
> > >>  > From: Jim Hendler (hendler@cs.umd.edu)
> > >>  > Date: Thu, Oct 31 2002
> > >>  >
> > >>  > *Next message: Peter F. Patel-Schneider: "Re: Guide: draft of
> > >>Oct 31 (goofy TranstiveProperty use)"
> > >>  >
> > >>  >    * Previous message: Smith, Michael K: "RE: Guide: draft of Oct 31"
> > >>  >    * In reply to: Ian Horrocks: "Re: Issue: Add hasValue to OWL Lite"
> > >>  >    * Next in thread: Jos De_Roo: "Re: Issue: Add hasValue to OWL Lite"
> > >>  >    * Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
> > >>  >    * Other mail archives: [this mailing list] [other W3C mailing lists]
> > >>  >    * Mail actions: [ respond to this message ] [ mail a new topic ]
> > >>  >
> > >>  >   ------------------------------------------------------------------------
> > >>  >
> > >>  > Message-Id: <p05111710b9e759013172@[10.0.0.16]>
> > >>  > Date: Thu, 31 Oct 2002 17:05:21 -0500
> > >>  > To: Ian Horrocks <horrocks@cs.man.ac.uk>, Dan Connolly <connolly@w3.org>
> > >>  > From: Jim Hendler <hendler@cs.umd.edu>
> > >>  > Cc: Deborah McGuinness <dlm@ksl.stanford.edu>, webont
> > >><www-webont-wg@w3.org>
> > >>  > Subject: Re: Issue:  Add hasValue to OWL Lite
> > >>  >
> > >>  > At 3:45 PM +0000 10/31/02, Ian Horrocks wrote:
> > >>  > >On October 29, Dan Connolly writes:
> > >>  > >>
> > >>  > >>  On Tue, 2002-10-29 at 18:34, Ian Horrocks wrote:
> > >>  > >>  >
> > >>  > >>  > The problem with adding hasValue to OWL Lite is that it wouldn't be
> > >>  > >>  > Lite any more. The lack of hasValue in Lite is, from an
> > >>implementation
> > >>  > >>  > point of view, the main thing that differentiates it from fast -
> > >>  > >>  > hasValue is very tough to deal with, and is responsible for pushing
> > >>  > >>  > the worst case complexity of reasoning in fast OWL from ExpTime to
> > >  > > >>  > NExpTime.
> > >>  > >>
> > >>  > >>  Could you unpack that a bit?
> > >>  > >>
> > >>  > >>  Could you give an example, maybe?
> > >>  > >
> > >>  > >I'm not sure. This isn't anything to do with reasoning techniques or
> > >>  > >specific examples, it is a fundamental property of the logic that
> > >>  > >basic inference problems (satisfiability, subsumption, entailment) are
> > >>  > >much harder when we add extensionally defined classes (which is what
> > >>  > >hasValue amounts to).
> > >>  > >
> > >>  > >If you want an intuition, it comes down to the loss of the tree(ish)
> > >>  > >model property. Without this property, it is very hard to devise
> > >>  > >decision procedures that work in a goal-directed way and that know
> > >>  > >when they are done.
> > >>  > >
> > >>  > >Ian
> > >>  >
> > >>  > Ian - the question arose at the Telecon as to whether this was true
> > >>  > for both the IF and the ONLY IF (i.e. hasValue -> X vs X -> hasValue)
> > >>  > -- that is, does saying "All Mexican restaurants serve Mexican food"
> > >>  > cause the problem if you're not expected to be able to say "all
> > >>  > places that serve Mexican food are Mexican restaurants"??
> > >>  >   -JH
> > >>  >
> > >>  > --
> > >>  > Professor James Hendler                           hendler@cs.umd.edu
> > >>  > Director, Semantic Web and Agent Technologies     301-405-2696
> > >>  > Maryland Information and Network Dynamics Lab.    301-405-6707 (Fax)
> > >>  > Univ of Maryland, College Park, MD 20742          240-731-3822 (Cell)
> > >>  > http://www.cs.umd.edu/users/hendler
> > >>  >
> > >>  >   ------------------------------------------------------------------------
> > >>  >
> > >>  >    * Next message: Peter F. Patel-Schneider: "Re: Guide: draft
> > >>of Oct 31 (goofy TranstiveProperty use)"
> > >>  >    * Previous message: Smith, Michael K: "RE: Guide: draft of Oct 31"
> > >>  >    * In reply to: Ian Horrocks: "Re: Issue: Add hasValue to OWL Lite"
> > >>  >    * Next in thread: Jos De_Roo: "Re: Issue: Add hasValue to OWL Lite"
> > >>  >    * Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
> > >>  >    * Other mail archives: [this mailing list] [other W3C mailing lists]
> > >>  >    * Mail actions: [ respond to this message ] [ mail a new topic ]
> > >>
> > >>  --
> > >>   Deborah L. McGuinness
> > >>   Knowledge Systems Laboratory
> > >>   Gates Computer Science Building, 2A Room 241
> > >>   Stanford University, Stanford, CA 94305-9020
> > >>   email: dlm@ksl.stanford.edu
> > >>   URL: http://ksl.stanford.edu/people/dlm
> > >>   (voice) 650 723 9770    (stanford fax) 650 725 5850   (computer
> > >>fax)  801 705 0941
> > >>
> >
> >
> > --
> > Professor James Hendler                                 hendler@cs.umd.edu
> > Director, Semantic Web and Agent Technologies   301-405-2696
> > Maryland Information and Network Dynamics Lab.          301-405-6707 (Fax)
> > Univ of Maryland, College Park, MD 20742        240-731-3822 (Cell)
> > http://www.cs.umd.edu/users/hendler

--
 Deborah L. McGuinness
 Knowledge Systems Laboratory
 Gates Computer Science Building, 2A Room 241
 Stanford University, Stanford, CA 94305-9020
 email: dlm@ksl.stanford.edu
 URL: http://ksl.stanford.edu/people/dlm
 (voice) 650 723 9770    (stanford fax) 650 725 5850   (computer fax)  801 705 0941

Received on Tuesday, 10 December 2002 13:54:15 UTC