RE: Semantics for Web Services Characterization

Steve,

I suspect this is a Pandora's box that may distract from my comments on
the charters - but I really can't resist because it's such a juicy
question that's always just under the surface in these discussions.

I think there's been a lot of confusion over whether OWL-S composite
processes describe orchestrations or choreographies. That is, do they
describe the web-service from the inside looking out, or from the
outside looking in. Looking back over the documentation (even when it
was daml-s) it seems clear that composite processes describe multi-step
"interactions or conversations between the user and services". More
importantly, these are 'glass box' views for the benefit of the client
so that they are better able to "make choices and provide information
conditionally". More recently, "It is important to understand that a
process is not a program to be executed. It is a specification of the
ways a client may interact with a service." The fact that OWL-S is not
an executable also provides the clue that it isn't a workflow
(orchestration) language, or even a fancy client-side scripting language
(ontologists should avoid programming language creep), but a loose set
of constraints that the client should adhere to in their interactions
with the service. The adoption of ontologies such as PSL to underpin the
semantics of OWL-S also reinforces this view of process as constraint.

Now, as far as I know, OWL-S describes only a _single_ service
interface, albeit comprising possibly many atomic (black-box) and
composite (glass-box) processes. When people speak of service
composition in the context of OWL-S they are talking about composing a
description of the service from these processes (service as process) all
of which are common to the same service instance, they are not speaking
about composing distinct services per-se. In conclusion, I think OWL-S
may actually be your missing link, defining the client's conversation
with (behaviour of) a single service.

Steve.

> -----Original Message-----
> From: Steve Ross-Talbot [mailto:steve@pi4tech.com] 
> Sent: 18 November 2005 11:23
> To: Battle, Steven Andrew
> Cc: Carine Bournez; public-sws-ig@w3.org; www-ws@w3.org
> Subject: Re: Semantics for Web Services Characterization
> 
> Bit late joining this discussion but I wanted to pick up on 
> something Steve said.
> Namely  "The proposal is to find out and demonstrate what 
> can't be achieved with the current Web Services 
> technologies.". One of the missing pieces as I see it is 
> defining the bahvior of a single service. In WS-CDL we define 
> the common and collaborative behaviour. We have need for an 
> end point view of behavior that is sound so that we can 
> project to it. It seems to me that such behavior is pretty 
> important for any understanding of semantics which might be 
> used for composition (statically or dynamically). Any 
> thoughts on this?
> 
> Kohei Honda, Nobuko Yoshida and Marco Carbone will have some 
> input on this as part of their work on formalisation 
> underpining WS-CDL. They have a precise description of an 
> end-point language that does just this. The chances are in 
> will become a working note from WS-Choreography WG so you 
> would be welcome to pick it up and run with it.
> 
> I'd be interested to hear more on what the ideas are around 
> behavioral semantics.
> Any comments?
> 
> Cheers
> 
> Steve T
> 
> 
> On 17 Nov 2005, at 23:04, Battle, Steven Andrew wrote:
> 
> >
> > Carine,
> >
> > I'm relieved to hear that the 'characterization' activity 
> is intended 
> > to consider a wide variety of solutions that build on WSDL. 
> It would 
> > be good to see this intent clarified in the charter 
> document. I know 
> > I'm beginning to sound like a stuck record but I believe 
> that WSDL-S, 
> > OWL-S and WSMO all build on WSDL and deserve equal mention. 
> > Alternatively, the focus should be entirely on the 
> use-cases in which 
> > case no specific technologies need be referenced.
> >
> > Steve.
> >
> >> -----Original Message-----
> >> From: Carine Bournez [mailto:carine@w3.org]
> >> Sent: 17 November 2005 17:37
> >> To: Battle, Steven Andrew
> >> Cc: public-sws-ig@w3.org; www-ws@w3.org
> >> Subject: Re: Semantics for Web Services Characterization
> >>
> >>
> >> Steve,
> >>
> >> Let me try to clarify the intent (it seems to me that 
> there is a deep 
> >> misunderstanding).
> >>
> >>> Given this rich context to draw on, it surprises me that the 
> >>> 'characterization' charter seems to limit itself almost
> >> exclusively to
> >>> "solutions like WSDL-S", which I read as invocation and a bit of 
> >>> discovery. This really isn't going to attract many relevant
> >> scenarios.
> >>
> >> The charter does not limit itslef to solutions like WSDL-S.
> >> The idea is to think about building a technology stack, 
> starting from 
> >> WSDL, adding some semantic extensions (generic enough to 
> be able to 
> >> build on top of these) and continue on those footprints. 
> The goal is 
> >> precisely to define the scope of what could be done (invocation? 
> >> discovery? more?).
> >> The proposal is to find out and demonstrate what can't be achieved 
> >> with the current Web Services technologies.
> >>
> >>> Given that the mission is to analyse "real-scale 
> applications", why 
> >>> eliminate composition, mediation, validation from the outset? For 
> >>> example, there's great opportunity here to work with the
> >> SWS 'mediation'
> >>> Challenge <http://deri.stanford.edu/challenge/2006/>
> >> organised by DERI
> >>> Stanford.
> >>
> >> Again, the charter does not exclude any of those, because those 
> >> particular "key points" should be determined by the group.
> >>
> >>> "The mission of the Semantics for Web Services
> >> Characterization Group
> >>> is to continue in the footprints of solutions like WSDL-S 
> and study 
> >>> the field of applications and identify key points that are not 
> >>> immediately solved using Web services technologies."
> >>>
> >>> could be changed to something like:
> >>>
> >>> "The mission of the Semantics for Web Services
> >> Characterization Group
> >>> is to study the field of applications addressed by
> >> technologies such
> >>> as WSDL-S, OWL-S, WSMO and SWSF and to identify key 
> points that are 
> >>> not immediately solved using Web services technologies."
> >>
> >> Restricting the scope to the fields that are already addressed by 
> >> existing technologies is IMHO a bad idea for characterization. The 
> >> goal is to derive the functionalities from the use cases, not from 
> >> the technologies developed in the area.
> >>
> >> I hope this helps.
> >>
> >> --
> >> Carine Bournez -+- W3C Sophia-Antipolis
> >>
> >>
> >
> >
> >
> 
>

Received on Friday, 18 November 2005 14:23:36 UTC