Re: Suitable OWL-S profile

> > [me]
> > In other words, if the service sells tickets, it should come with a
> > description that says ... This service sells tickets.  Inferring this
> > from the existence of an output is unnecessary and not even sound.

> [Katia Sycara]
> I certainly did not mean to imply that this should be inferred (ie
> that the service sells tickets). So, what would be the description
> that says that the service sells tickets? Certainly, one way of
> representing this description is for the service to have tickets
> as ( one of ) is outputs.

I liked it better when you said that the client's owning of the
tickets would be an effect of doing business with the service.
I don't see how in general you could interpret "Y is an output"
to mean "A new Y exists and the client owns it."  Outputs are just
pieces of information, aren't they?  Even if you did want to allow
physical outputs, you would still have lots of extra information about
the details of the transfer of ownership.  If I buy a house from a
web realtor, is the house an output?  An input?  Suppose I buy a corn
future?  Don't I have to spell out somewhere what is that I own and in
what sense?  Given the need for all that extra information, it doesn't
seem worthwhile to complicate the concept of outputs to avoid having
to say "Effect: the client owns Y."

> As to csp's question, yes it is very possible that we will need 
> ontologies of human travel behavior. In past research (on human
> tasking of intelligent agents for internet-based tasks such as
> buying tickets), we have called these "task domain fragments", ie
> ontologies that characterize tasks in a particular domain, such as
> travel. For example, one would like to know that travel vehicles
> include ships, cars, trains, airplanes, that for business travel,
> besides the travel vehicle tickets one also needs a hotel etc.

So if you drove a dialogue system or form-filling program by the
domain-fragment ontology, Charlie Abela would have what he's asking
for? 


                                             -- Drew

-- 

                                         -- Drew McDermott
                                            Yale University
                                            Computer Science Department

Some see the glass as half empty, some as half full.  Surely the
truth lies somewhere in between.

Received on Friday, 19 November 2004 03:05:23 UTC