RE: OWL-S: Parameter modeling

Bijan,

Thanks for your detailed answer. I think it time to do my homework now. I will come back to you once I understand better the issue with second-order logic. 

The broken link for Cyc is (one line):

http://www.cyc.com/doc/tut/ppoint/SecondOrderPredicates_files/frame.htm

Best regards
 
Stephane Fellah
Web Chief Architect
 
PCI Geomatics
490, Boulevard St Joseph
Hull, Quebec
Canada J8Y 3Y7
Tel: 1 819 770 0022 Ext. 223
Fax 1 819 770 0098
Visit our web site:  www.pcigeomatics.com
 


-----Original Message-----
From: Bijan Parsia [mailto:bparsia@isr.umd.edu] 
Sent: Friday, July 02, 2004 1:25 PM
To: Stephane Fellah
Cc: public-sws-ig@w3.org
Subject: Re: OWL-S: Parameter modeling

On Jul 2, 2004, at 2:07 PM, Stephane Fellah wrote:

> Bijan,
>
> Thanks for your detailed response. I understand now why I am in  
> OWL-Full. Basically the approach I suggest, is using second-order  
> predicate (meta-property).

Well. you are using second-order *syntax*. You really don't want to go  
true second order. It's not only undecidable, but incomplete (in  
general), and fairly difficult to work with. First order logic is  
challenging enough :)

Second order syntax has a lot of appeal, as you've pointed out. OWL  
Full is designed have that appeal without having the pain of full  
second order.

However, this can cause other difficulties. I strongly recommend  
checking out Hilog. You also will find some nice, accessible discussion  
in:	
	http://www.ihmc.us/users/phayes/SCL-december.html

>  Note that Cyc is also using working second-order logic reasoner using  
> second-order predicate. Ref.  
> http://www.cyc.com/doc/tut/ppoint/SecondOrderPredicates_files/ 
> frame.htm
> What is wrong with their approach ?

Sorry, that page isn't rendering for me.

> I think when it comes to model function or process, adding the support  
> of second-order predicate in reasoners will simplify considerably the  
> description of process or functor. The question is whether this  
> addition of second-order predicate (a small subset of second-order  
> predicate logic) would keep the reasoner decidable and can compute in  
> a finite time.

Most probably not, if you introduce true second order variables.

However, this is exactly why OWL Full, HiLog, and common logic take the  
second order syntax, first order semantics approach.

However, this causes them to *deviate* from the standard mapping into  
first order logic. For example, instead of classes being first place  
predicates, they are triples. (a rdf:type C). To get back the *class*  
semantics, you need an axiom (or many), if (a rdf:type C) then Ca.  
Getting the right axioms, and proving that they get it right, etc. etc.  
is tricky.

I am working on some other ways to accomplish some of this.

> Why should we stick religiously to DL ?

It's not about religion, it's about accomodating various user  
communities, tools, existing reasoners.

> Not being an implementer of inference engine, I am asking whether it  
> is a huge task to add this second order predicate extension in  
> existing DL reasoner or not ?

Depends on how you do it. The OWL Full way, I'd say yes. It's all about  
trade offs.

Personally, I wouldn't do this for readability. Have a surface language  
for that.

Cheers,
Bijan Parsia.

Received on Friday, 2 July 2004 14:38:34 UTC