Re: Draft OWL-S Process technical

> [Donal Murtagh]
> See below for my two cents:
> 
> . 
> (P.3,4)
> It's not clear to me how to interpret the example precondition. It is
> initially described as:
> 
> "in order to send the number of a certain credit card to a web agent,
> one must know what its number is"
> 
> It's not clear here whether "its" refers to the credit card number or
> the web agent's number. 

It refers to the credit card's number.

> Subsequently, it is stated that "Verifying that
> the process is feasible requires computing the available credit limit",
> but there was no mention of credit limit in the initial description of
> this precondition. Could you explain how this precondition *should* be
> interpreted?

Some gibberish crept into my usually crystalline prose.  It should
read --

  Verifying that the process is feasible requires retrieving the
  credit-card's 16-digit number, which is then associated with the
  variable \tstt{\#Num}.  We call such a parameter a \emph{local}
  parameter.

> 
> Also, my understanding is that SWRL is the language which will most
> commonly be used to specify Expressions, so why not use that instead of
> DRS/Kif. The fact that SWRL is more readable than DRS/Kif IMHO is
> another reason to prefer it for explanatory purposes over DRS/Kif.

I'm not sure what you mean by DRS/Kif; they're very different
syntaxes.  My own preference is for Kif above DRS and SWRL, which are
very similar.

I think I avoided SWRL in the context of this particular example
because I don't know how to express 

(Know-val-is (credit-card-num ?cc) ?num)

in SWRL.  Suggestions would be most welcome.

> (P. 8,9) 

References to page numbers are rather 20th-century, don't you think?
It's not like I actually ever printed the text out.  It would be more
helpful to give some nearby character strings as context.

> Where is the binding of the variables ?creditCard and ?purchaseAmt
> specified?

They are bound in the <hasInput> specs at the top of the AtomicProcess
declaration.  I wasn't consistent in my capitalization, so sometimes
they're written as, e.g., ?creditCard instead of ?CreditCard.  They
should all be capitalized.

> What exactly do the various axioms in the expressions represent? For
> example is (credit-limit ?creditCard) a function call which must
> evaluate to true, or a fact which must exist in the KB?
> 
I'm assuming that, in essence, assertions like this are in the KB:

   (current-value (credit-limit cc20) (cash-in dollars 6303))

which would then match with the query

   (current-value (credit-limit ?CreditCard) ?CreditLimH)

Many thanks for your comments.

                                             -- Drew

-- 
                                             -- Drew McDermott
                                                Yale University CS Dept.

Received on Tuesday, 29 June 2004 14:25:06 UTC