ISSUE-5: n-ary datatypes - lots of choice, no guidance

6) Various decidable subsets are possible, but no basis for making choices


Assuming the design is fixed to permit referring to some n-ary 
datatypes, then there are many different choices for n-ary datatypes 
over which it is in principle possible to reason.

Carsten points out:
[[
1. the integers with addition, equality, and equality to constants

2. the integers with multiplication, equality, and equality to constants

Both of them are useful and decidable (1. can be reduced to integer
programming or Presburger arithmetic, and 2. to Skolem arithmetic) but
together they are not because of Hilbert 10.
]]

further possibilities are to restrict oneself to only finite datatypes, 
such as float, int, ... which being finite are necessarily decidable 
(although practical optimization may be challenging).

Carsten suggests:
[[
Would it be a solution to give a couple of (alternative) datatypes
that OWL1.1 could support, and also to point out the combinations
that cause troubles? I feel that being liberal here is fine, because
in the end it boild down to what you reasoner supports anyway.
]]

This seems to me to miss the point:
- a W3C specification is provided to provide interopertation between 
reasoners
- a W3C specification is provided to allow interoperation between data 
from different sources
- without guidance from a specification, the reasoner vendors have all 
decided to implement the same datatype group, the one that contains no 
datatypes: this is the one that is ready for standardization.

I feel that the analysis provided by Turner and Carroll demonstrates 
conclusively that we do not know how to provide such interoperation, and 
hence the area of n-ary datatypes is either unsuited to standardization 
or is being put forward for standardization prematurely.
In either case the solution for this WG is to support only 1-ary 
datatypes, and to postpone the issue awaiting further research.

Received on Monday, 12 November 2007 20:25:12 UTC