W3C home > Mailing lists > Public > public-owl-dev@w3.org > October to December 2010

Re: All humans love (all) cats

From: Bijan Parsia <bparsia@cs.man.ac.uk>
Date: Thu, 7 Oct 2010 15:59:40 +0100
Cc: pklinov@cs.man.ac.uk, public-owl-dev@w3.org
Message-Id: <AEF8E6EF-1ED0-4B04-A6FD-7E481F33372C@cs.man.ac.uk>
To: Adrian Walker <adriandwalker@gmail.com>

On 7 Oct 2010, at 15:12, Adrian Walker wrote:

> Hi Bijan,
> 
> Thanks for your  most erudite exposition of some aspects of OWL.
> 
> You wrote
> 
>> ...where's your proof that your representation above is equi-satisfiable with the various OWL versions? If it is not, then you haven't solved it. The burden of proof is, obviously, on you.
>> 
> It's likely that at least some of the versions of OWL are incomparable  with Executable English / Internet Business Logic.

That's not the question.

Your rule version doesn't capture the first order version.

The challenge is to get an equisatisfiable encoding of the first order version. That is what you repeatedly fail to even attempt or acknowledge.

The OWL version is "complex" because it's trying to encode a complex feature *faithfully*. Whether one *needs* that faithfulness is an orthogonal question.But then if one settles for an approximation, it's easier to do in OWL (at least, extended with DL Safe rules).

>  (i.e. There are overlapping but distinct Vennn diagrams for expressivity)
> 
> However my question still appears to be on the table: what do we buy with the added conceptual complexity** of OWL over EE/IBL rules?

No, it isn't. You can try to make that the question, but it's not a respectable move.

> As an expert on OWL who is also well read on the earlier work on stratification, you can surely point us to further crisp examples that will work in OWL but may pose a challenge for EE/IBL rules.

Hijacking this thread (yet again) is not respectable either. 

> Thanks in advance,   -- Adrian
> 
> 
> **See for example from the current discussion:
> 
> "I feel rather miffed by the fact that such a simple and mundane-looking FOL statement requires such an excruciatingly  complicated workaround in order to be represented in OWL2."

Pointer?

So, two points, one about you and one about this statement.

About you: You've not shown that you can encode this simple and mundane-looking FOL statement in EE/IBL. We've already shown (I think reasonably, if suggestibly) that your version is not correct. (e.g., I feel confident that your system/version doesn't support contrapositive reasoning with classical negation)

Until you put forth an actual solution (with at least a prima facie argument that it actually captures the statement) nothing is required of anyone else. In other words, ante up.

About that statement: Be miffed, but it's a bit silly to be. It's obviously the case that you can't encode all sorts of "mundane" but arbitrary first order statements in OWL. That's the price of computational reasonableness. Engineering is about tradeoffs. Now, *this construct* happens to be encodable in OWL, which, as Markus has argued in print, makes it a reasonable candidate for a direct implementation (or, at least, a preprocessing based implementation). So, if it's important enough, it could be added to OWL 2 (or 2.1). What *cannot* be done is make it so all "nice" FOL statements are directly encodable in OWL, for obvious reasons.

But (back to you), obviously that's also the case for IBL. FOL is far more expressive.

Cheers,
Bijan.
Received on Thursday, 7 October 2010 14:58:37 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:58 GMT