skolemization proposal

I think the apparent disagreement on skolemization is based on 
misunderstanding.

I believe that this is the only obstacle to resolving ISSUE-3.

====

Here is a proposal concerning skolemization

In our instructions for test cases we add the following observation:

[[
Certain legitimate implementation techniques for OWL reasoners are 
unable to easily compute some or all of the entailments which include 
blank nodes in their conclusions. Any such test failure does not reflect 
on the conformance of the reasoner to the OWL Semantics, or to 
conformance labels such as OWL Consisteny Checker.
]]

i.e. the formal semantics should remain unchanged, but any misreading of 
the recommendation to suggest that OWL Reasoners must implement 
entailment, and must implement entailment with existentials, should be 
explicitly headed off. There is no such requirement in OWL 1.0, and no 
one has argued for such a requirement in OWL 1.1

In the OWL 1.0 group we, IIRC, explicitly decided to only have 
consisteny checking as the conformance label, and not owl reasoner, 
minded that they were different methods of implementation and we did not 
wish to mandate any particular one.

IIRC that was at the WebOnt manchester F2F .... Manchester is a fun place!!

Jeremy

Received on Monday, 10 December 2007 15:57:04 UTC