Re: model-theoretic semantics for DAML-ONT

just to further support Richard's comments,
thx for the work on another semantics description.
I think this is further progress towards community understanding of the
language.

I think having the informal specification is useful for some of our
audience.   As this semantics gets further refined, what I (and I think a
number of readers ) would like to see is for
the more informal statements in the original draft (refinined if necessary) to
be maintained
and then be augmented  with the more formal description filling in the
details.

It would also be great to continue to point out where each of our current
drafts differ (as peter mentioned on cardinality).   For now, I think
identifying differences and attempting to gain consensus on one intended
meaning for complete reasoners would be best.  (if and when we diverge on
meanings, I think we want to have compelling reasons and also we want to keep
a list of the places so we can explain the differences to users.  Right now on
this issue, I think we just need a little more discussion on cardinality (also
recognizing that cardinality is about to be augmented in daml-ont to include
domain specific cardinalities.))

We all know daml-ont is likely to have a number of different incomplete
reasoners using it so that will mean it essentially ends up with a number of
different semantics.  If the differences can remain due to different choices
in what inference rules to drop, I think we will maintain better
interoperability and ease of use.

Deborah

Richard Fikes wrote:

> > I put together a draft of a model-theoretic semantics for DAML-ONT as a
> > counterpoint to the axiomatic semantics recently submitted to this list.
> >
> > Comments are welcome.
>
> I am pleased to see a start on another description of a semantics for
> DAML-ONT.  This initial description seems to be rather informal and
> incomplete in important ways.  I encourage you to make it more formal
> and more complete by covering all of the DAML-ONT spec.
>
> A couple more general comments come to mind:
>
> The same points that you have made about the axiomatic semantics
> requiring the reader to understand the syntax and semantics of KIF can
> also be made here.  That is, you have used set theory (which one?),
> symbols such as "<=", terms such as "disjoint", "=", etc. without
> defining them or providing references to the definitions you intend.
>
> Many of the statements in the semantics (e.g., the ones about
> restrictions and qualifications) seem to be English prose statements of
> the axioms that are in the axiomatic semantics.  Is there some reason to
> prefer the informal prose statements over the formal axioms?
>
> It seems you have avoided using constructs like "holds" primarily by
> stating the properties of the various DAML-ONT terms and constructs in
> English.  How is that better?
>
> I look forward to the next version of this semantics which I hope will
> be more formal and more complete.
>
> More importantly, I hope that these two semantic descriptions of
> DAML-ONT will continue to bring to the surface issues in the design of
> DAML-ONT that need further attention and clarification, and thereby
> support the DAML-ONT community.
>
> Richard

--
=======================
Deborah L. McGuinness
Associate Director and Senior Research Scientist
Knowledge Systems Laboratory
Stanford University
Stanford, CA 94305
dlm@ksl.stanford.edu
voice  650 723 9770
fax 650 725 5850

Received on Thursday, 16 November 2000 14:34:17 UTC