Re: does SWS standardization need to wait for a rule language?

Hi Jacek and all,

preconditions and effects are quite important, and I think SWS should
provide a simple and compact way to represent them. I did some work to
explore how SPARQL can be used for this purpose, and I think it provides
a good solution.

The idea (in a few lines) is the following. A SPARQL CONSTRUCT query
form can give a compact representation of preconditions and effects:
- the query graph pattern in the WHERE clause defines constraints that
must be satisfied in order to invoke the service (i.e. preconditions);
- the resulting RDF graph (built according to the CONSTRUCT template)
gives a description of the state of the world after the service
invocation (i.e. effects).

I proposed this approach for OWL-S (at the OWL-S workshop at ESWC 2007),
 and you can read details in my paper [1]. The approach is generic, and,
although I've described it for OWL-S, it is not actually tied to any
specific OWL-S construct, and I think it's reusable and effective.

Any comment is welcome.

Best Regards
marco

[1]
http://www.ai.sri.com/OWL-S-2007/final-versions/OWL-S-2007-Sbodio-Final.pdf

Jacek Kopecky wrote:
> Hi all,
> I've thought about this in the context of the latest discussions about
> the next steps of SWS standardization:
> 
> A big part of SWS descriptions is ontologies, and we have OWL for that.
> But another significant part is various logical expressions
> (preconditions, effects) which can't be expressed in OWL. There's the
> RIF working group that I expect to give us a useful rule language for
> that, but what about the scheduling?
> 
> I have a feeling that we may need to wait until a useful RIF spec is in
> Last Call at least before new work starts that expects to use it. Or
> could maybe SPARQL be used somehow?
> 
> Any opinions?
> Jacek
> 
>

Received on Thursday, 12 July 2007 07:35:03 UTC