Re: [OWL-S] Proposal for Making Progress on Process Modeling esp. Preconditons and Effects

(www-rdf-rules added to followups, in subsequent posts, I'll only  
follow up to www-rdf-rules)

On Jan 15, 2004, at 12:42 PM, Michael Kifer wrote:

>> Bijan Parsia <bparsia@isr.umd.edu> wrote:
[snip]
>> I'd be interested in a CTL based axiomization of OWL-S. Are you doing
>> one for SWSL?
>
> I am planning to do it this week as part of my SWSL action item.

Great, looking forward.

>
>> What features would SWRL need to support CTL? How easy is it to  
>> combine
>> OWL and CTL?
>
> These are two orthogonal things.

Hence the two questions :) Actually, they aren't quite so distinct as  
SWRL (at least last I checked) is a proper superset of OWL-DL. There  
are issues with certain extensions of predicate calculus discussed in  
the Description Logic Programs paper.

> CTL extends predicate calculus, while OWL
> (at least, OWL-DL) is a subset of predicate calculus.

Sure.

>  Of course, it is
> a different issue whether one can make a useful combination out of the  
> two.

Er...I have to take this to mean that there might be *more* FOL that  
one would want than either OWL-DL or SWRL. I.e., if the extensions are  
compatible with all of FOL, then there's no problem adding them to a  
subset. The question is whether the extended subset is at all  
interesting. (E.g., is Concurrent Transaction Propositional Logic  
interesting from a practical SWS point of view.)

> In my view, the most important thing(s) about DLs are the subsumption
> algorithms.

Really? Subsumption is typically reduced to satisfiability. That is,  
the core reasoning service is satisfiability. This can be used, in the  
normal refutation manner, to support relatively arbitrary entailments  
(there are some issues since you don't typically have direct negation  
of property assertions, but see Ian and Peter's paper on reducing OWL  
Entailmen to DL satisfiability:  
http://www.cs.man.ac.uk/~horrocks/Publications/download/2003/ 
HoPa03b.pdf)

> It is not clear how and where you are going to use subsumption
> when it comes to process models (i.e., sequencing, parallel  
> composition,
> loops, etc.).

See above. There are classification algorithms, but they are mostly  
optimization oriented, that is, trying to avoid as many subsumption  
tests as possible. But the standard subsumption test is:
	To see if C isSubclassOf D
	Take the kb, add (C & ~D)
	Check the satisfiability of the resultant kb.
		If unsatisfiable, then the subsumption holds.
		If not, then it doesn't hold.

There's nothing magical about subsumption.

[snip]

>> I think Benjamin's example (in figure 1) can be handled in OWL-S, with
>> simpleprocesses, though perhaps not as elegantly.
>
> You can (almost) always avoid defaults for any particular example by
> blowing up specifications and incorporating exceptions directly into  
> the
> spec. The problem with this is that this is inelegant, makes it hard to
> write the specs, and makes the specs non-compositional. New exceptions
> must be re-incorporated into the specs, etc. The reason why
> object-oriented programming became so successful is because it offers a
> way to avoid all this complexity.

I'll wait until I've given it a shot to see what you think. I'm not  
clear that I *would* be avoiding non-mon, per se. Notice I said in  
*OWL-S*, not in OWL -- much of the semantics of OWL-S descriptions is  
not expressed in OWL.

Cheers,
Bijan Parsia.

Received on Thursday, 15 January 2004 13:18:56 UTC