Re: A possible way of going forward with OWL-R unification (ISSUE-131)

On Jul 11, 2008, at 2:28 AM, Alan Ruttenberg wrote:

> On Jul 10, 2008, at 4:20 PM, Boris Motik wrote:
>
>> "The rules from Section 4.3 can be applied to arbitrary RDF  
>> graphs, in which case the produced consequences are sound but not
>> necessarily complete."
>
> One thing to consider with this last bit, is that there is  
> issue-117 and discussion at the F2F had leaned towards saying that  
> non-entailments in OWL-R would not be sanctioned. Thus "complete"  
> would need to be qualified - the entailments might be complete in  
> the sense that no others are sanctions, but incomplete with respect  
> to a more expressive language.

Just for the record, at the time, I understood that discussion *not*  
to be about additional RDF graphs, but to be about some entailments  
wrt the syntactic fragment...that is, as strictly analogous to the  
situation with, e.g., the treatment of finite owl:Things in DL vs.  
OWL Full, *NOT* as about arbitrary other graphs.

Perhaps I was the only one, but I certainly would have been, er, more  
"vigorous" in the discussion in my inimitable way if I had understood  
it to be going along these lines.

I don't see the problem with saying that a rule engine with the OWL-R  
rules is a sound but incomplete OWL Full engine (just as an OWL DL  
reasoner is a sound but incomplete (we think) OWL Full engine)) and a  
sound and complete OWL-R engine. That seems to cover the situation  
exactly, using standard lingo, and understandably.

Cheers,
Bijan.