- From: David Martin <martin@AI.SRI.COM>
- Date: Tue, 27 Apr 2004 12:49:12 -0700
- To: Michael Kifer <kifer@cs.sunysb.edu>
- Cc: "Battle, Steve" <steve.battle@hp.com>, "Public-Sws-Ig@W3. Org" <public-sws-ig@w3.org>
Michael Kifer wrote: > Steve, > According to you, a postcondition is nothing but a logical consequence of > the frame axioms. This is not very useful. > > Why would you want to specify a postcondition like that if it can be > derived by a reasoning system? Because of the limitations of the reasoning > system? > > In my view, a postcondition should be viewed as a constraint on the > after-state of an action, if the effect of the action is non-deterministic. What about my original example(s) - where what I want to express is something about the relationship between the input(s) and output(s). For instance, I want to say something like: (1) interest_based_on_daily_compounding(InRate, InBalance, OutInterest) where the predicate interest_based_on_daily_compounding is defined in some standard accounting ontology. Would you be comfortable calling (1) a postcondition? Or does postcondition, for you, have to do with an "after-state" that is persistent? Thanks, David > > > --michael > > > Battle, Steve writes: > >>Here's another way of thinking about this - essentially from the situation >>calculus. An effect describes things that are true _because_ of an action, >>whereas, a postcondition describes things that are true _following_ an >>action. >> >>Not everything that is true following an action is true because of it. A >>small example : If I add item A to my shopping trolley, then the effect is >>that "item A is in my trolley". If I then add item B to my trolley, then the >>effect is that "item B is in my trolley". Now, it's reasonable to assume >>that "item A is in my trolley" remains true because nothing I've done claims >>to effect the truth of it. So given that "item A is in my trolley" is a >>precondition of the "add item to trolley" action then "item A is in my >>trolley" is a reasonable post-condition, but not an effect. A lot of good >>work has gone into working out what post-conditions are reasonable given the >>effects (see the 'frame problem'). >> >>Steve >> >> >>>-----Original Message----- >>>From: public-sws-ig-request@w3.org >>>[mailto:public-sws-ig-request@w3.org]On Behalf Of Charlie Abela >>>Sent: Wednesday, April 14, 2004 6:26 PM >>>To: Public-Sws-Ig@W3. Org >>>Subject: Re: Effects in OWL-S >>> >>> >>> >>>With all these ideas going on about preconditions and effects >>>in OWL-S it is >>>quite difficult to capture the general idea of how to define >>>actual effects >>>in WSs. >>> >>>I had the impression that an effect was something that will >>>become true when >>>the WS has executed but that also brought some changes to the >>>world, but now >>>there is talk of making use of post-condition instead. >>>Actually from the >>>readings that I found, these two words seem to be used >>>interchangeably, >>>depending on the research context, and thus I always presumed >>>that they are >>>somewhat synonymous. >>> >>>What are the views of the OWL-S ppl on this? Cause with all >>>these somewhat >>>radical changes being proposed it is quite difficult to get people to >>>actually make use of these ontologies. Will there every be a >>>stable set of >>>OWL-S ontologies? >>> >>>Some time back there was a long discussion on the topic; I guess some >>>clarification is now due. >>> >>>Regards >>> >>>Charlie >>> >>> >>> >>> >>> >>> >>> >> >> > >
Received on Tuesday, 27 April 2004 15:48:24 UTC