W3C home > Mailing lists > Public > www-ws@w3.org > September 2003

Re: `: why can't describe the semantic of DAML-S by Description Logic

From: Bijan Parsia <bparsia@isr.umd.edu>
Date: Tue, 2 Sep 2003 15:12:19 -0400
Cc: <www-ws@w3.org>
To: "lsp" <lsp@is.pku.edu.cn>
Message-Id: <5F7DC294-DD79-11D7-A704-0003939E0B44@isr.umd.edu>

On Tuesday, September 2, 2003, at 02:55  PM, lsp wrote:

> Yes, I had thought to resort to the expressive description logic
> language called ALCIreg[1], which corresponds directly to Propositional
> Dynamic Logic. I believe it can fully describe the process model if
> taking web service as role in DL.

It can only describe propositional conditions (pace some exceeding 
cleverness). This is very limited.

It would require rather a different representation than the current 

> As for the matchmaking problem, we can regard the ServiceProfile as
> Concept and
> still apply subsumption  reasoning for matchmaking.

Uh...well...this isn't actually responsive to the issues involved. I 
have no idea what it is to "regard the ServiceProfile as Concept", 
precisely. I recommend the Horrocks paper as it deals with some of the 
trickery of using subsumption for matchmaking.

> I'm trying to describe the semantic of DAML-S by ALCIreg, and this 
> would
> be my doctoral research proposal, I'd like to know the feasibility of
> this approach.

In one respect, feasible but somewhat trivial. In other either 
infeasible or very very very hard.

Representing if-then, repeat, sequence, etc. Easy, but done to death :) 
"Hey, look, let's map the control constucts into their corresponding, 
well, control constructs in PDL!" "Yeah, and?" The tricky bits lie 

Again, if you could show a class of processes that  1) are "natural" 
for Web service based compositions and 2) are significant (i.e., a 
large, likely set of common Web service related tasks), and 3) are 
exactly describable by PDL, then I think you have a winner.

And, let me add, I'm prolly the *most* enthusiastic and hopeful person 
connected to DAML-S about this approach :)

> Thanks for any feedback.

Other approach I've toyed with is trying to describe petri nets 
directly in OWL. There's a little work that I know of about this (by 
the Racer guys, IIRC).

But really, pace cleverness, there are some fairly hard expressive 
limits here.

Hope this helps.

Bijan Parsia.
Received on Tuesday, 2 September 2003 15:09:57 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 23:05:12 UTC