Formal Semantics of OWL + RDF + SPARQL + SWRL

Peter --

A thought about "those were not the results I expected" and SPARQL...

One of the bad things about the various implementations of classical SQL is 
that they have divergent semantics -- that is, they can produce different 
results given the same query and data.

For combined use of OWL + RDF + SPARQL + SWRL it seems even more important 
that there should be a simple, accessible model theory (analogous to [1]) 
that implementers of the combined languages can adhere to with a reasonable 
amount of effort.  (This is argued in more detail in [2]).

Does such an animal exist, or is there perhaps work in that direction?

Thanks in advance for pointers.    -- Adrian


[1]   Backchain Iteration: Towards a Practical Inference Method that is 
Simple Enough to be Proved Terminating, Sound and  Complete, A. Walker. 
Journal of Automated Reasoning, 11:1-22.

[2]  Understandability and Semantic Interoperability of Diverse Rules Systems
www.w3.org/2004/12/rules-ws/paper/19

(With apologies if this note has some html in it.  My email program is a 
bit independent about outgoing format)



INTERNET BUSINESS LOGIC (R)
Online at www.reengineeringllc.com
Shared, community use is FREE.

Adrian Walker
Reengineering
PO Box 1412
Bristol
CT 06011-1412 USA

Phone: USA 860 583 9677
Cell:    USA  860 830 2085
Fax:    USA  860 314 1029



At 12:48 PM 12/5/2005 -0500, you wrote:

>Hmm, adding SPARQL into the mix is indeed adding a significant new piece 
>of the
>puzzle.  The SPARQL spec is still in flux, I believe, and some readings of it
>may support the behaviour you are experiencing.
>
>peter
>
>
>From: Andrea Splendiani <andrea@pasteur.fr>
>Subject: Re: Question about SWRL semantics
>Date: Mon, 5 Dec 2005 18:41:14 +0100
>
> > Ok,
> > I just forgot one point.
> > I was testing it querying through SPARQL,
> > I guess anyway SPARQL should se an "inffered set of relations", o that
> > is not an issue. Isn't it ?
> >
> > The semantics you describe is the one I expected.
> >
> > best,
> > Andrea Splendiani
> >
> > Il giorno 05/dic/05, alle 18:31, Peter F. Patel-Schneider ha scritto:
> >
> > >
> > > From: Andrea Splendiani <andrea@pasteur.fr>
> > > Subject: Question about SWRL semantics
> > > Date: Mon, 5 Dec 2005 18:15:14 +0100
> > >
> > >> Hi,
> > >> I['m] asking this here but feel free to redirect me to some more
> > >> specific
> > >> resources if it is the case.
> > >>
> > >> I was using SWRL rules in Protégé and I have noticed that SWRL does
> > >> not
> > >> "view" the relations that can be inferred from OWL.
> > >>
> > >> Is this a feature of SWRL semantics ?
> > >> Is is just undefined and "implementation specific" ?
> > >> Is it due to the inability of the reasoner (Pellet) to infer too much
> > >> about instances ?
> > >>
> > >> best,
> > >> Andrea Splendiani
> > >
> > > If you have a KB (however this is formed) that includes both SWRL
> > > rules and OWL
> > > axioms then a complete SWRL reasoner should "take note" of both the
> > > rules and
> > > the axioms, as specified in the SWRL documentation.
> > >
> > > So, for example,
> > >
> > >     Class(Student Person)
> > >     Class(Rock Immobile)
> > >     Individual(John type(Student))
> > >     Implies(Antecedent(Person(I-variable(x)))
> > > Consequent(Rock(I-variable(x))))
> > >
> > > entails
> > >
> > >     Individual(John type(Immobile))
> > >
> > >
> > > I do not know why you are not obtaining the results you expect, but it
> > > seems to
> > > me that you are expecting the results that are provided by the SWRL
> > > semantics.
> > >
> > > Peter F. Patel-Schneider
> > > Bell Labs Research
> > >
> > >
> > >

Received on Tuesday, 6 December 2005 02:11:05 UTC