RE: Proposed response to Martin Merry, HP

>
> Jeremy - that's simply not true - there are people who have
> implemented every feature in combination - I'm not sure what more
> than that you can mean by a "research hypothesis" -- I think you're
> just misunderstanding the situation or else I simply don't understand
> you.  However, for every test case you've proposed we seem to have
> people who think they can implement it - so I have no clue what you
> mean by a "theoretical construct"
>   -JH
>

I am not sure what to make of this either - I may have misunderstood the
situation or you may have done or we may have different concerns.

We struck the concept of a "complete DL consistency checker" on the grounds
that we did not think there would be one.

So for Lite we already have (I believe) complete consistency checkers, for
Full we know that there cannot be such things, and for DL we believe these
not to be commercially practical. I think that this is sufficiently relevant
to the new user trying to work out how to get into OWL to make it in to the
Guide in some way, and don't find the text as proposed as spelling out these
different levels of practical reasoning adequately.

i.e. if you confine yourself to OWL Lite then hopefully you can take your
data from one OWL Lite system with a complete reasoner to another without
major rework.

If you confine yourself to OWL DL then such portability is less expected -
while all DL features may be implemented we do not expect an implementation
of all of them all at the same time, (at least not "complete").

In OWL Full then expect less portability than that.

However, on the plus side, even in Full any reasoning a conformant OWL
reasoner does will be consistent with any other reasoning a different
conformant OWL reasoner has done.

Jeremy

Received on Friday, 13 June 2003 10:01:43 UTC