Re: DAML-S ProcessModel

Drew,

You suggest that if there was a core ontology with terms such as "owns", 
"pays", etc., that goals could be stated in a formal way. And then -- 
presumably -- reasoners could determine the best way to meet those goals.

Isn't this going the wrong direction?

Terms such as "owns" and "pays" get their meaning be pointing into a 
shared interaction model. Let's say that this is the "everyday 
interaction model for commerce" that exists in the actual world. (Does 
it? Is there just one? Is is a cultural artifact? Lets suppose that 
those are orthogonal questions.) Then the goal directed reasoners will 
manipulate this model in order to arrive at their results.

But -- as I understand it -- DAML-S is proposing a way to present formal 
interaction models directly, without this long circuit out into the 
world of markets and such that has historically given meaning to the 
"owns" and "pays" kind of terms.

In the DAML-S way of doing things, you just reason with the interaction 
models that you are presented with, directly. So the core ontology terms 
become sort of ancillary. Perhaps they are only needed for metadata, or 
for documentation purposes, to describe what happened.

Jeff


Drew McDermott wrote:

>   The requester specifies its "desired outcome", e.g., 
>   "I want to buy a book of J.R.R. Tolkien". 
>   In the cyberspace it means that "an invoice for a book 
>   of Tolkien is delivered to the requester". 
>
>Why not go all the way and say that the formal meaning of the goal is 
>(informally!) "I own a book by Tolkien (that I didn't already own)"?
>
>   The "desired outcome" must be expressed in a common 
>   generic conversation language. 
>   (As far as I know there are no means to express 
>   explicitly the "desired outcome" in DAML-S)
>
>DAML-S provides for goal statements, but as yet does not specify a way
>to express them.  An appendix (which I wrote) suggests a syntax
>somewhat compatible with RDF, but I am sure genuine extensions to RDF
>will eventually be endorsed by the W3C.  In any case, the real issue
>is, as you said, a common language, that is, a common vocabulary or
>"ontology."  There will probably never be a vocabulary agreed upon by
>everyone, but it's certainly reasonable to assume that there will be a
>core ontology with terms such as "owns," "pays," and such.
>
>  
>

Received on Tuesday, 20 May 2003 13:26:34 UTC