Re: Situation Calculus Expressiveness question

Hi Sheila --

Some comments about your question (below)....

1.  I understand that your axiom is intended as a simple test 
case.  However, in a blocks world you would also want to say also that a 
robot cannot pickup A if it is already holding A.

2.  I'd guess that state transitions in the Situation Calculus can be 
encoded in most respectable AI planning systems.  However, control and 
efficiency are likely issues.

3.  If SWRL is to be extended to cover the Situation Calculus, the 
underlying inference engine may have to deal with issues of nonstratified 
negation-as-failure (E.g. see comment 1 above).  That is, the rules 
describing the Situation Calculus will likely turn out to contain recursive 
loops in which a closed world negation occurs.  Stable model semantics aims 
to deal with this, but possibly introduces problematic "don't know" results.

4.  Just for fun, a slightly augmented version of your axiom can be run by 
pointing Netscape 7 or Mozilla to www.reengineeringllc.com , and choosing 
the demo "ForSheilaMcIlraith".

Hope this helps.        Cheers,  - Adrian



                                            INTERNET BUSINESS LOGIC

                                              www.reengineeringllc.com

Adrian Walker
Reengineering LLC
PO Box 1412
Bristol
CT 06011-1412 USA

Phone: USA 860 583 9677
Cell:    USA  860 830 2085
Fax:    USA  860 314 1029





At 03:42 PM 12/17/03 -0500, you wrote:

>Hi all,
>
>The following question arose from an OWL-S [1] Coalition discussion
>regarding development of a process model for Web services (WS).  OWL alone
>is not sufficiently expressive to capture all and only the intended
>interpretations of a WS process model, and we're investigating whether
>SWRL may be.
>
>To this end, we are asking whether  we can axiomatize a situation
>calculus [2] domain theory in SWRL.  The situation calculus
>calculus is a first-order logical language for reasoning about
>action and change.  It has proven sufficiently expressive for
>axiomatizing a Web service process model, and we wondered whether
>such a process model could be expressed in SWRL.  If not, can
>SWRL be extended to axiomatize a situation calculus domain theory?
>
>
>To this end, the following is an example of an axiom we would like
>to encode:
>
>Forall x. Forall s.
>   holding(x,do(a,s)) iff
>     [(a=pickup(x)) V (holding(x,s) & (a neq putdown(x))]
>
>Legend:
>- do is a function that maps actions (a) and situations (s) into
>    new situations (s).
>- V is "or"
>- neq is "not equal"
>- iff is "if and only if"
>
>
>This axiom says
><a robot is> holding x in the situation resulting from
>performing action a in situation s (i.e., "do(a,s)") if and only if
>- the action was "pickup(x)"
>OR
>- <the robot was> holidng x is situation s, and the action was
>   not "putdown(x)"
>
>
>[1] OWL-S is the an OWL ontology for services
>     http://www.daml.org/sevices
>
>[2] The situation calculus is a first-order language for reasoning
>     about action and change.  It can originally be credited to McCarthy
>    and Hayes.  Reiter was central in extending the language and renewing
>    interest in its use.
>
>     "Knowledge in Action: Logical Foundations for Specifying
>         and Implementing Dynamic Systems"
>      Raymond Reiter, 2001, The MIT Press, Cambridge, Mass.
>
>
>Thanks,
>Sheila
>
>
>----------------------------------------------------------------------
>Sheila McIlraith
>Department of Computer Science
>University of Toronto
>Toronto, Canada M5S 3H5
>
>sheila@cs.toronto.edu
>http://www.ksl.stanford.edu/people/sam/

Received on Wednesday, 17 December 2003 17:36:14 UTC