Re: "FAST" OWL and "Greater" OWL

Ian,

I have a different take on this. Indeed if we were presented with Greater
OWL _alone_ I'd share your concerns. Rather we are presented with two
options:

1) one perhaps more restricted, better characterised and with known software
techniques for implementation.
2) the other looser, less characterized and indeed a more researchy field.

The risk is that if Greater OWL is proven untenable, then we are left with
Fast OWL. Sounds fine to me. What is the real problem with that except that
it requires a bit more work to develop _both_ model theories? (on the other
hand if Pat is willing to take on this work then I for one am in no position
to complain...) Indeed it would be more efficient to go with one or the
other, yet that is the price for compromise. Indeed a 50 or so member WG is
hardly the most efficient way to get something done. Indeed I'd prefer that
our conference calls spend alot more time actually debating the technical
merits of technical issues rather than spending time deciding where to hold
future f2fs and the seemingly endless amounts of other overhead issues that
burden us. The benefits of a large group are that we can get many different
folks to work together (as best possible). In the very IETF spirit of "let a
thousand flowers bloom" I think Pat's approach is a VERY GOOD THING(tm).

That said, I see no reason why we ought not debate the merits of Peter's
proposal -- well, I can't be the judge of which MT is better but I'd like to
hear a _rational_ analysis of the merits of Peter's proposal vs. "Greater
OWL" for example.

It seems as though if we get enough model theories for OWL we can open
another issue to decide whether OWL accept the intersection or union of the
entailments licensed by each of the various model theories as the actual
entailments for OWL.

Jonathan


>
> Several points regarding your latest document(s).
>
> 1. I have been arguing for a long time that trying to resolve the
> layering problem (or any issue come to that) by inventing a new logic
> is not a sensible thing for a standardisation working group to do. And
> make no mistake, "Greater OWL" is a new logic about which, by Pat's
> own admission, we currently know very little. Peter has been doing a
> great job of pointing out just some of the immediately obvious
> pitfalls. Pat is apparently willing to wager a glass of scrumpy that
> he can fix (some of) the problems. Are we willing to bet the future of
> OWL against a glass of scrumpy?
>
> To borrow Jim Hendler's drug analogy, the drug we are being offered
> here has been knocked together in a few months, we know little about
> its possible effects (except by analogy with similar drugs), and it
> has never been subjected to clinical trials. I think that these facts
> would make a material difference to most people's decision as to
> whether or not they wanted to risk taking the drug.
>
> 2. It is disingenuous, or at least misguided, to denote the more
> restricted form of layering as "Fast OWL", to depict the restrictions
> as strange and unnecessary, and to imply that their only justification
> is to facilitate the use of some weird kind of reasoning system.
>
> In fact, "Fast OWL" could better be described as "First Order OWL":
> the restrictions are just those required to keep OWL within the
> standard first order framework, i.e., where classes directly
> correspond to formulae with one free variable, subsumption is
> implication, etc. Unlike "Greater OWL", the languages within this
> framework, up to and including standard FOL, have been extensively
> studied and are very well understood.
>
> Staying within this framework imbues OWL with a simple and very well
> understood semantics, and allows applications to use wide range of
> reasoning engines ranging from DLs through to full FOL
> systems. Moreover, it would facilitate the future extension of the
> language, i.e., to include the expressive power of full FOL.
>
> 3. If we really need a document describing how OWL can be embedded in
> RDF with semantic constraints such that OWL entailment and RDF
> entailment coincide, then I would like to draw the WGs attention back
> to Peter's proposal of Aug 28 last [1] that details a much simpler way
> to do this than "Greater OWL".
>
> Regards, Ian
>
> [1]
http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/att-0136/01-part
>
>

Received on Tuesday, 24 September 2002 20:29:10 UTC