Re: N-ary Data Predicates

Excerpts from Bijan Parsia's message of Tue Jun 17 10:05:48 -0400 2008:
> 
> Hi folks,
> 
> Now that Easy Keys are in the "bake" stage (i.e., we've reach  
> consenus (with appropraite reservations) and have spec text in  
> specs), the next most mature proposal to fill a capability hole is n- 
> ary data predicates.
> 
> A reminder of use cases:
>     http://www.w3.org/2007/OWL/wiki/N-ary_Data_predicate_use_case
> 
> The current draft of a proposal (needs a tart up to match new specs):
>     http://www.w3.org/2007/OWL/wiki/N-ary_Data_predicate_use_case
> 
> For at least linear inequations, we have implementation, indeed,  
> commercial experience in the form of Racer Pro.
> 
> Finally, the only general objector to n-ary per se (HP) has withdrawn  
> that objection. So I think we're no longer in need of reaching  
> consensus on the basic facility, but on exactly what we'll spec and  
> require.

FWIW, Evan Wallace (I'm not speaking for him here!) and I are participating in
the W3C's Incubator Group on Product Modeling, which is an OWL-related effort.
I'm very new to PM, generally, but it's already clear that PM has *tons* of
use cases and requirements for a quite robust n-ary data predicate design in
OWL2.

They define classes like Facade, which is (making this up!) a subclass of
BuildingPart that has a height twice as long as its width.

They also want to be able to say things like a particular kind of stairway has
a width that is some proportion of its enclosing structure -- as I understand
things, this is beyond what anyone proposes for OWL2, and I mention it only to
give an idea that there are real customer needs here.

Again, I don't speak for Evan, but I do encourage him to backfill some of my
sloppy tech details here with some actual content. :>

Cheers,
Kendall Clark