RE: Internet/Distributed Computing using HTTP/POST: Bridge semantic W eb and Web services under the same Internet protocol

Ahh - I like to see a good flame war :)
Steve.

> -----Original Message-----
> From: Bijan Parsia [mailto:bparsia@isr.umd.edu] 
> Sent: 20 January 2006 14:08
> To: Battle, Steven Andrew
> Cc: public-sws-ig@w3.org; Shi, Xuan
> Subject: Re: Internet/Distributed Computing using HTTP/POST: 
> Bridge semantic W eb and Web services under the same Internet protocol
> 
...
>> OWL-S cannot compose services, only processes that 
> > ultimately break down into atomic processes that correspond to WSDL 
> > operations. In other words, OWL-S only addresses compositions of 
> > actions that can be performed at a _single_ service interface.
> 
> I don't believe that at all. Or, to try to make the latter 
> part of your last sentence true, since my service interface 
> might call out to other service, and OWL-S describes how I do 
> that, OWL-S certainly can....
>

I'd like to believe you, really I would. I should know better, but just
for fun I'm going to pull out my "BLUFF" card. But other than in the
'motivating tasks', the OWL-S documentation only ever discusses single
services, not service composition.

It comes down to the old nutshell about what OWL-S is talking about -
behind the scenes orchestration, or the choreography of the visible
interaction with a service. My evidence, "a composite process is not a
behavior a service will do, but a behavior (or set of behaviors) the
client can perform by sending and receiving a series of messages". This
doesn't sound like orchestration to me.

> > It can't
> > describe, for example, how you can buy a book on Amazon 
> then sell it 
> > on eBay because these are two different services.
> 

My understanding is that any atomic process has only two participants,
'TheClient' and 'TheServer' (see below). My understanding is that the
latter is implicitly the server provider agent for whatever service is
being described.
Where does OWL-S distinguish between different service providers (e.g
Amazon and eBay)? I just don't see it in the ontology.

<owl:Class rdf:about="#AtomicProcess">
    <rdfs:subClassOf>
      <owl:Restriction>
        <owl:onProperty rdf:resource="#hasClient"/>
        <owl:hasValue rdf:resource="#TheClient"/>
      </owl:Restriction>
    </rdfs:subClassOf>
    <rdfs:subClassOf>
      <owl:Restriction>
        <owl:onProperty rdf:resource="#performedBy"/>
        <owl:hasValue rdf:resource="#TheServer"/>
      </owl:Restriction>
    </rdfs:subClassOf>
</owl:Class>

> Do this.
> 
> I mean, we have constructive existence proofs!
> 

I'd like to see how the kind of service composition you describe can be
expressed in the OWL-S process model (not by the back-door in the
grounding, or in an extension). Then I'll admit I'm wrong and administer
suitable self-flagellation.

> [snip]
> > You say that, "the service aggregation process does not provide the 
> > semantic meaning of the services and such aggregations 
> should not be 
> > exposed to the service requesters who are waiting for an answer to 
> > their request". I think you've misunderstood the subtlety of OWL-S 
> > process modelling. An OWL-S process does not expose 
> anything going on 
> > behind the service interface, rather "it is a specification of the 
> > ways a client may interact with a service" see 
> > <http://www.daml.org/services/owl-s/1.1/overview/#5>. In 
> other words, 
> > every step of a composite process has to be performed by the client.
> [snip]
> While I think that's way too strong (and have argued against 
> what I see as an artificial restriction in the text), even 
> granting this undermines your earlier assertion. (Plus, it 
> can't be that I *perform* the action, in most non-technical 
> uses of perform and many technical ones...the client 
> *invokes* it...and with a simple process I don't see why it 
> could transfer control).
> 

Granted - the performance of an atomic process is an invocation. The
main thrust is that "a composite process is not a behavior a service
will do, but a behavior (or set of behaviors) the client can perform by
sending and receiving a series of messages." 

I don't read it as an artificial restriction but as a clear statement
about the nature of OWL-S.

> Cheers,
> Bijan.
> 
>

Received on Friday, 20 January 2006 16:33:31 UTC