W3C home > Mailing lists > Public > public-sws-ig@w3.org > October 2004

Re: ambiguous match and business transaction

From: Drew McDermott <drew.mcdermott@yale.edu>
Date: Fri, 1 Oct 2004 03:00:54 +0000 (GMT)
To: public-sws-ig@w3.org
Message-ID: <E1CDDfh-0008Oo-Hx@frink.w3.org>


> [Daniela CLARO]
> >- Concerning the service Profile .... I read that the service profile
> permits agents or other programs to check unambiguously the matching
> >between a requested WS and a advertised WS, only with input and output > (I
> don't really understand how the condition and effect works)  >... But
> imagine you can found 2 services that have the input / output, but don't > do
> the same thing ... like this example : You have 2
> >inputs, 2 Numbers and 1 output, also a number. The service A do input1 > 
> input2 while the service B do input1 - input2. How a matching >algorithm > can
> make a difference between this 2 services (because they have the same
> input/output) ?

Let's go back to your comment "I don't really understand how the
condition and effect works."  Conditions and effects contain the
answer to your question. Suppose we represent service A thus:

Action: (request-to-A ?input1 ?input2 - (Literal Integer)
Value: (?res - (Literal Integer))
Effect: (know-val (+ ?input1 ?input2) ?res)

and B thus:

Action: (request-to-B ?input1 ?input2 - (Literal Integer))
Value: (?res - (Literal Integer))
Effect: (know-val (- ?input2 ?input1) ?res)

[Some mail-reader interference has made it unclear exactly how your
example went, but I trust this version is more or less what you
wanted.]

In other words: service A sends back a number such that the recipient
then knows it is the sum input1 + input2, whereas B sends back a number
such that the recipient knows it is the difference of input1 and
input2.

The challenge is for an agent in possession of this model to use it to
figure out that it should send a request to A when it wants to add two
numbers. 


                                             -- Drew


-- 
                                             -- Drew McDermott
                                                Yale CS Department
Received on Friday, 1 October 2004 03:01:25 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 16 March 2008 00:10:57 GMT