W3C home > Mailing lists > Public > www-rdf-rules@w3.org > January 2004

Re: Expressiveness question

From: Sheila McIlraith <sheila@cs.toronto.edu>
Date: Wed, 28 Jan 2004 10:36:36 -0500
To: Ian Horrocks <horrocks@cs.man.ac.uk>
Cc: pat hayes <phayes@ihmc.us>, www-rdf-rules@w3.org, daml-process@bbn.com
Message-ID: <Pine.GSO.4.58.0401281032090.2092@dvp.cs>


Ian,

An answer to your question below.

On Tue, 27 Jan 2004, Ian Horrocks wrote:

> On January 27, Sheila McIlraith writes:
> >
> > Ian,
> >
> > This is very helpful and interesting.  <Though I need to work through
> > an example in more details to understand all of the implications.>
> >
> > Regarding the question of whether the disjunctive literals in FOL would be
> > encoded as classes or properties in OWL, my sense is that actions (e.g.,
> > a=pickup(x)) would be encoded as classes.  I'll think about whether
> > fluents in the situation calculus (predicates, indexed by the situation
> > term, whose truth value can changes as a result of an action) could be
> > encoded as classes as well.
>
> It isn't obvious to me how you can capture the relationship between a
> and x using a class, but I will wait to hear more.

Think of an action "a" (e.g., "pickup"), as being analogous to a process
in the OWL-S service ontology, and the variable x, being analogous to an
input parameter.  In the original DAML-S, processes (e.g., "pickup") were
modelled as classes, and their input (e.g., "x") as properties.  In the
most recent OWL-S, processes are now instances of classes, but the same
relationship holds.


> If it is the case
> that you can encode one or other of the disjuncts as classes, then I
> believe that you can capture the iff using the rewriting trick that I
> described.

Great.


Sheila


>
> Ian
>
> >
> > Sheila
> >
> >
> >
> > On Tue, 27 Jan 2004, Ian Horrocks wrote:
> >
> > > On January 25, Sheila McIlraith writes:
> > > >
> > > >
> > > > Hi Pat,
> > > >
> > > >
> > > > On Tue, 20 Jan 2004, pat hayes wrote:
> > > >
> > >
> > > [...]
> > >
> > > > > holding(x, do(a,s)) IMPLIES ((a=pickup(x)) OR (holding(x,s) )
> > > > >
> > > > > isn't, and isn't ever likely to be stateable in any rule language.
> > >
> > > But given that in SWRL combines rules with OWL, we get something much
> > > more powerful which may allow us to state more that in normal rule
> > > languages. E.g., if the disjunction in the head of the rule included a
> > > unary predicate:
> > >
> > > Body IMPLIES P1(x) OR P2(x)
> > >
> > > then we would be able to state it in SWRL because we can rewrite it as
> > >
> > > Body AND NOT P2(x) IMPLIES P1(x)
> > >
> > > SWRL allows us to use (NOT P2) as a predicate (or we could use OWL to
> > > assert that the class NOT-P2 as equivalent to the negation of the
> > > class P2).
> > >
> > > Whether or not this kind of trick would work for the rule Pat wrote
> > > would depend on how (a=pickup(x)) and (holding(x,s)) are encoded: if
> > > they are encoded as binary predicates (OWL properties), then it seems
> > > unlikely that we can express it in SWRL as it would amount to
> > > providing property negation, and Uli Sattler has managed to convince
> > > me that we (almost certainly) can't express property negation in SWRL.
> > >
> > > Regards,
> > >
> > > Ian
> > >
> --
> [To unsubscribe to this list send an email to "majdart@bbn.com"
> with the following text in the BODY of the message "unsubscribe daml-process"]
>
Received on Wednesday, 28 January 2004 10:37:14 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 2 March 2016 11:10:15 UTC