- From: Juergen Zimmer <jzimmer@inf.ed.ac.uk>
- Date: Thu, 02 Dec 2004 10:57:54 +0000
- To: Manshan Lin <lmshill@gmail.com>
- CC: public-sws-ig@w3.org
Hi, this is just my 2 cents... Manshan Lin wrote: >hi all, >I've just read all the discussion about applying AI planning technique to SWS. >I notice that most traditional planning algorithm are based on first >order logic, which is based on close-world assumption. And I also >notice that in order >to achieve true automatic composition of web service, we must tackle >the problem of planning under description logic, which is based on >open-world assumption. > > > that is true and Drew has also pointed that out in his paper " *Estimated Regression Planning for Interactions with Web Services" * >The question is : >When using DL to describe world state, what adaptions should >traditional AI planning techniques make? >For example, >TBOX: EffectA = intersectionOf(EffectB,EffectC) >ABOX: a individual EffectA >Then we can choose an operation that achieves EffectA or we can choose >two operations (one achieves EffectB and the other achieves EffectC). >It's a little like adding some common rules >(in this case EffectB(x) and EffectC(x)->EffectA(x)) >to traditional planning domain. When EffectB and EffectC are not >atomic concepts, the situation becomes more complex. How to handle >this kind of things in planning algorithms? > > I've experimented with the (rather old) PRODIGY planner for SWS composition and supposedly it allows to add axioms like this to planning domain descriptions. Manual at: http://www-2.cs.cmu.edu/afs/cs.cmu.edu/project/prodigy/version4.0/pub/manual.ps.gz So your example would probably along the lines of (Inference-Rule intersectionClass (params <x>) (preconds ((<x> Effect)) (and (type <x> EffectB) (type <x> EffectC))) (effects () ((add (type <x> EffectA))))) Basically, inference-rules are like operators except that they don't occur in the final plan. However, I havn't really worked with this feature so far... One problem is that Prodigy's type system only allows single inheritance. :( Personally, I think that planning is not really the right approach to SWS composition. You find many alternative approaches on http://staff.um.edu.mt/cabe2/research/projects/swsc/planning.htm Situation calculus seems appealing but, of course, there is no usable implementation and I don't know much about automation of it... Cheers Juergen > > > >Best regards! > >Manshan Lin >Email: lmshill@hotmail.com;lmshill@gmail.com >Affiliation: School of Computer Science and Engineering, the South >China University of Technology >Phone: (+86)13711287277 >2004-12-02 > >
Received on Thursday, 2 December 2004 10:57:58 UTC