Re: Punning and the "properties for classes" use case (from public-owl-dev)

On Nov 2, 2007, at 12:36 PM, Jeremy Carroll wrote:

> By default, in a WG trying to provide a relatively small revision  
> on an established specification, the alternative to making a change  
> is to make no change.
>
> This is HP's preferred option here: i.e. no change from OWL 1.0:  
> punning prohibited in OWL DL (at least in the RDF form); when a uri  
> is used, this takes you into OWL Full, and each URI denotes one thing.
>
> I point to charter text in favour of this position:
> [[
> For each new feature, if there is doubt or a perceived problem with  
> respect to this issue, the guideline should be to not include the  
> feature
> ]]

This one is with regards to backwards compatibility. That sneaky bit  
that Peter pointed out from the OWL 1.0 now makes that the backwards  
compatibility question a little more complicated, and we might  
discuss, rather, whether a repair is needed so that the RDF and  
abstract syntax aren't at variance. (personally, I can't believe that  
it went through in the first place!)

> and
> [[
> The existing compatibility between OWL DL and OWL Full should be  
> preserved
> ]]

Here the question is, what exactly "compatibility" means. I've a  
pretty good message from David Turner about this, that I should put  
on the wiki, but it doesn't seem like it's a simple question. Do you  
have a short articulation of what you think the test of this is?

By making these points, I not trying to suggest that we resolve this  
issue on technicalities - I'd rather we didn't do that. It's just to  
point out that the application of those rules will not be  
straightforward. I'd much rather have the discussion move towards  
figuring out some solution that can approach meeting all party's needs.

What did you think about the suggestions I started to make about  
additional structural specification that more strongly make the point  
that a URI is intended to denote one resource?

Also, I'd like to understand the reasoning behind Ian's assertion

> Name separation is required, however, if Fast OWL is to be embedded  
> in RDFS in such a way as to be semantically compatible with Large OWL.

  (BTW, what's Fast OWL and Large OWL?)


-Alan

Received on Friday, 2 November 2007 17:01:19 UTC