Re: Automatic Web Service Composition

Dear Nicola,

I think it is important to be clear just what one may mean by automatic 
discovery and composition of "Services". I think it is equally 
important that one is clear about what one may mean by "Semantic" Web 
Services as opposed to Web Services.

As usual in all of this the devil is in the detail.  When we talk about 
Web Services do we ground at WSDL2.0 or earlier? When we talk of 
"Semantic"  Web Services what do we mean by the term "Semantic"?

If we look at the move towards Web Services it is clear (even now) that 
it is happening as we speak. WSDL2.0 promises better interoperability 
(as does SOAP1.2) but this does not incorporate semantics in any 
classic sense. If you have behavioralist tendencies then WS-BPEL with 
it's abstract process definition provide some answers to this and if 
you widen the net to multiple services working in a peer-to-peer 
environment then the work of W3C Choreography certainly captures the 
external observable behavior of such collaboration; which I think is 
the more interesting case.

When one talks of composition if it is taken in the context of gaining 
better execution (price and/or performance) then the service 
description at a WSDL level may not need to change (that is many 
candidate services may provide the service) but the level of service 
that each may provide may differ. So when we talk about composition do 
we mean the generation of new code or do we mean the leveraging of 
existing services?

If we mean composition through the leveraging of existing services then 
the work of the WSD working group (WSDL2.0 to be) and the work of 
Choreography working group may have considerable relevance to your 
research. On the one hand there may be many things that one can infer 
from a WSDL2.0 description of a Web Service along with any RDF meta 
data attached to it. On the other hand the behavioralist approach taken 
by the Choreography working group and its roots in process algebras may 
provide you with interesting semantic properties in terms of the 
contract that one may wish to exist between cooperating Web Services.

An area of considerable interest to many users at the moment is how to 
capture SLA/Policy requirements against a services so that these sorts 
of properties may be used in service location which may predicate 
service composition in a choreographic sense.

Cheers

Steve T

On 2 Feb 2004, at 09:03, Nicola Dragoni wrote:

> I'm beginning to work on my Ph.D. Thesis which concerns the automatic 
> discovery and composition of Semantic Web Services.
> I know works which use AI planning techniques to address the automatic 
> composition problem. Does anyone know of work that addresses this 
> issue and that does not use AI planning techniques?
>
> Thanks,
>     Nicola
>
> ----------------------------------------------------
> Dr. Dragoni Nicola
> Ph. D. Student in Computer
> Science
> e-mail: dragoni@cs.unibo.it
>            nicoladr@libero.it
> web: www.cs.unibo.it/~dragoni
> ----------------------------------------------------

This email is confidential and may be protected by legal privilege. If you are not the intended recipient,  please do not copy or disclose its content but  delete the email and contact the sender immediately. Whilst we run antivirus software on all internet emails we are not liable for any loss or damage. The recipient is advised to run their own antivirus software.

Received on Monday, 2 February 2004 21:39:19 UTC