W3C home > Mailing lists > Public > www-ws@w3.org > July 2004

Re: DAML-S/OWL-S 0.9 Beta Release and New DAML-S Web Pages

From: Drew McDermott <drew.mcdermott@yale.edu>
Date: Wed, 7 Jul 2004 11:45:51 -0400 (EDT)
Message-Id: <200407071545.i67Fjpwl003632@pantheon-po04.its.yale.edu>
To: www-ws@w3.org


>    [Martin Gulich]
>    I have a few questions I am not able to answer = completely by reading
>    the OWL-S specification. I would be very grateful if you can = answer
>    any of them. If this is not an appropriate mailing list for such
>    questions = please direct me to one.

If you are basing your questions on version 0.9, you may want to check
version 1.0.  Owl-S is a moving target.  There is even a version 1.1
beta, but I don't know if it's publicly accessible.
> 
>    First, who is responsible for the execution of = a service that
>    consists of a composite process? Is the client that is invoking the
>    service responsible for looking up each atomic process and invoking it
>    = separately, that is, the client will need to function as some kind
>    of manager for = the service execution?

Yes.  (It's not clear what alternative there is.  If a sequence of
actions by the client is necessary to interact with a service, then
presumably the client will manage that sequence.)

>    Secondly, it is impossible to dynamically = compose a composite
>    process "ad hoc", right? For this task some kind = of agent planning
>    system is needed to dynamically compose a sequence of services = and
>    execute it?

That's my assumption.  Someone else may have a different paradigm in mind.

>    And as a follow-up question; is it possible to = dynamically compose
>    such a sequence out of whole services (found in a register) as = well
>    as examining an advertised service and compose a sequence out of some
>    (but = not necessarily all) of its atomic processes? I mean, that is
>    what you do when = statically composing a new WS out of atomic
>    processes and publishing it, isn=B4t = it?

It seems to me that a dynamic (= run-time) service-interaction planner
would not care where the services came from that it was stitching
together.

>    And finally, do you have any references to = the absolutely newest
>    research in the areas of capability based matching and dynamic
>    composition of semantic web services?

Check out recent proceedings of the ISWC: the International Semantic
Web Conference, e.g.:

  Dan Wu, Bijan Parsia, Evren Sirin, James Hendler, and Dana Nau 2003
  Automating DAML-S web services composition using SHOP2.  {\it Proc.
  Second Int'l Semantic Web Conference (ISWC2003)}, Sanibel Island,
  Florida

-- 
                                   -- Drew McDermott
                                      Yale Computer Science Department
Received on Wednesday, 7 July 2004 11:45:53 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:25:46 GMT