Re: ISSUE-5: n-ary datatypes - unimplemented incomplete design

Carsten Lutz wrote:
> Hi Jeremy,
> 
> I believe it is not true that there are no reasoners supporting n-ary
> datatypes.  For example, RACER does. And it works rather well.
> 
> greetings,
>         Carsten
> 


Hmmm, referring to the racer reference manual 1.9.2-beta p61

you might mean that:

RacerPro supports reasoning over natural numbers (N), integers (Z), 
reals (R), complex numbers (C), and strings.


This is not about support for OWL 1.1 n-ary datatypes. of those 
datatypes, only the natural numbers, integers, and strings are OWL 1.1 
base datatypes, there is no support in OWL for reals or complex numbers.

So, while I am impressed that the racer reasoner combines some aspects 
of numeric and qualitative reasoning, I would like to have clarification 
as to its bearing on OWL 1.1

Example issues that spring to my mind:

the fahrenheit to celsius stuff works because they are usign the Reals - 
in a domain like the decimals, floats or doubles, this all looks much 
harder.

The comment that nonlinear polynomials forces the domain to be the 
complex numbers seems like a non-trivial design choice. For example, the 
existential theory of the reals (nonlinear polynomials over the reals) 
is decidable.

Jeremy

Received on Monday, 12 November 2007 22:15:47 UTC