W3C home > Mailing lists > Public > public-owl-wg@w3.org > July 2008

Re: ISSUE-126 (Revisit Datatypes): A new proposal for the real <-> float <-> double conundrum

From: Bijan Parsia <bparsia@cs.man.ac.uk>
Date: Fri, 4 Jul 2008 17:38:14 +0100
Message-Id: <5DFB055C-ABE6-49EB-A8D8-066A4DDB3CC3@cs.man.ac.uk>
Cc: "Boris Motik" <boris.motik@comlab.ox.ac.uk>, "'OWL Working Group WG'" <public-owl-wg@w3.org>
To: Alan Ruttenberg <alanruttenberg@gmail.com>

On 4 Jul 2008, at 16:08, Alan Ruttenberg wrote:

> As I've mentioned, I've done some prospecting on his issue with  
> people who would be users of this feature. They considered the  
> counting/discreteness issue completely uninteresting "Why would  
> anyone want to do that?".

I'm not sure what the take away here is supposed to be. I presume  
they've never had to consider the problem before. I find that many  
users don't have a clear grasp of how they might use a feature (this  
cuts both ways, of course). With computational numerical methods, I  
think its especially important to be cautious in innovating. It's  
fairly difficult to predict the effects.

For example, suppose at some point we want to introduce n-ary  
predicates over floats and doubles. Whether we do inexact or  
interval, it's possible that having a continuous values space will  
produce *wrong* answers.

> Consider perhaps a nonstructural restriction disallowing counting  
> on floats.

That's one possibility. But I think it's not implementationally  
harder, in point of fact. It's not *good*, but we have to face the  
challenge anyway.

> I definitely not a fan of features no one will implement and they  
> are, in any case, at risk of being removed automatically by the two  
> implementations criterion.

I don't believe there is any such automatic removal. OWL1 went to  
without sound and complete recommendations. OWL Full has never had  
remotely plausible or complete implementations. Does this mean that  
OWL Full will be automatically removed from our spec?

> But personally, as an advocate for the inclusion of these types,  
> I'd rather have mincardinality of a float/double range that  
> corresponds to nonzero measure real interval be always satisfiable.

I would be very hesitant to get on board with that without a more  
extensive examination of the consequences,  semantically speaking.  
One thing that the EasyKey experience showed me is that intuitions  
can be really dangerous here.

Cheers,
Bijan.
Received on Friday, 4 July 2008 16:46:04 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:42:05 UTC