W3C home > Mailing lists > Public > public-sws-ig@w3.org > November 2004

RE: Relation between OWL and OWL-S

From: Daniela CLARO <daniela.claro@eseo.fr>
Date: Wed, 24 Nov 2004 09:32:12 +0100
To: <public-sws-ig@w3.org>
Message-Id: <20041124083215.869DD155D@harley.eseo.fr>

Great David...
Thank you very much...
I have already created an example based on owl-s examples, and it works
perfectly. And actually I have created the subclassof Profile, Process...

I am seeing right now my service as actually an instance of process.owl
ontology, and as you said, I will take a look in sameIndividualAs. It is
very clear right now...

And about the discovery process, I probably need to read more papers, so I
will take a look into pub-archive.html, because I understood something but
right now I think that some "pieces in my puzzle" is missing. 

Thank you very much,
Daniela


-----Message d'origine-----
De : David Martin [mailto:martin@AI.SRI.COM] 
Envoyé : mercredi 24 novembre 2004 09:12
À : daniela.claro@eseo.fr
Cc : public-sws-ig@w3.org
Objet : Re: Relation between OWL and OWL-S

Getting back to Daniela's original quesions, there are some straightfoward
answers -

Daniela CLARO wrote:

> Hi all,
>  I would like to know what is actually the relation between OWL-S and OWL?

OWL-S is an ontology represented in (the language) OWL.  This ontology
contains a bunch of class and property definitions.

A description of a particular service in OWL-S is (by and large) a set of
instances of the OWL-S classes and properties.

> How do I represent this relation in OWL-S? 

You don't have to do anything.  The relation between OWL and OWL-S is
already captured in the OWL-S declarations (in Service.owl, Process.owl,
Profile.owl, and the other ontology files).  It's captured by all the
declarations that say, for example, that "AtomicProcess is an OWL class',
"hasInput is an OWL property", and so forth.


> I will be more clear...for example, suppose that I have a service 
> called AirplaneCompany that searches for a ticket beased on some input 
> parameters and it belongs to a service composition called Travel.
> 
> - How can I do automatic discovery in this case? 

I won't try to answer that here, but there are some papers here:
   http://www.daml.org/services/owl-s/pub-archive.html
and elsewhere that talk about discovery.  In general, discovery and
matchmaking can take advantage of description-logic techniques like
subsumption, but also in general we've found that *pure* subsumption doesn't
do well enough; various modifications and extensions to it may be needed.

> I've read that the
> automatic discovery in OWL-S is based on its input and outputs 
> parameters, so we can make a distinction between the services.
> 
> But actually, how can I say that my AirplaneCompany service is a class 
> in OWL model?

Well, an OWL-S "service" is something that accomplishes a specific request,
like "ReserveTicket", so AirplaneCompany is more likely to be a name for a
collection of services rather than a service.  So I will answer using the
name ReserveTicket.


Actually an ReserveTicket service would be an *instance* of the Service
class (which is defined in Service.owl).

  Where Can I put this relation in OWL-S?

You simply make an OWL declaration that says "ReserveTicket is an instance
of owl-s:Service".  In other declarations you will also declare something
like "ReserveTicketProfile is an instance of "owl-s-profile:Profile" and
"ReserveTicketProcess is an instance of "owl-s-process:Process".

You can see examples of this here:

   http://www.daml.org/services/owl-s/1.1/examples.html

  And also, where can I say
> that AirplaneCompany is the same service that AircraftCompany, as we 
> do in OWL using equivalenceClass like:
> 
> <owl:Class rdf:ID="Aircraft">
>     <owl:equivalentClass rdf:resource="urn:eg#AirPlane"/>
> 
> These relations exist? Or in OWL-S we can not say that? 

Yes, you can say things like that, but since an OWL-S service is an
*instance*, you would use the OWL property sameIndividualAs, rather than
equivalentClass.

Regards,
David
Received on Wednesday, 24 November 2004 08:32:17 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 16 March 2008 00:10:58 GMT