ISSUE-120 (broken OWL 1 Full semantics): Fixing the inconsistency of OWL 1 Full will break perfect backwards compatibility

ISSUE-120 (broken OWL 1 Full semantics): Fixing the inconsistency of OWL 1 Full will break perfect backwards compatibility

http://www.w3.org/2007/OWL/tracker/issues/

Raised by: Michael Schneider
On product: 

[This issue is primarily meant to inform the WG. I suggest to close the issue as resolved, after the WG has taken note of the problem.]

It has originally been a desired aim to have complete backwards compatibility of OWL 2 Full with OWL 1 Full, where "backwards compatibility" means that every OWL 1 Full entailment is also an OWL 2 Full entailment.

The semantics of OWL 1 Full have turned out to be inconsistent due to bugs in the semantic conditions for unionOf, intersectionOf, and oneOf axioms.

Since OWL Full has a model-theoretic semantics, inconsistency means that for arbitrary RDF graphs G1 and G2 the entailment "G1 |= G2" holds (because G1 cannot be satisfied). 

If the mentioned bugs are corrected in order to make OWL 2 Full consistent (which is easy), there will afterwards exist RDF graphs G1 and G2, for which the entailment "G1 |= G2" does /not/ hold anymore.

So, formally, OWL 2 Full will not be backwards compatible with OWL 1 Full. However, it can still be achieved that OWL 2 Full will be backwards compatible with the /corrected/ version of OWL 1 Full.

Received on Sunday, 20 April 2008 23:32:00 UTC