Dear friends,

  I think the paper accurately describes the tensions surrounding
BioPAX and I like the proposals to extend OWL with one caveat: I would
not propose them as extensions of OWL, but as a separate layer.

  The extensions are very useful and easily implemented in principle,
but deeply clash with OWL principles. As the authors understand, they
do not freely blend with OWL reasoning. The suggested restrictions do
not behave like established OWL restrictions:

  - Usually in OWL, class restrictions are inherited to subclasses,
but the suggested restriction of no direct instances is not.

  - Usually, OWL reasoning can continue at any point, but some
proposed extensions are intended to be used after established OWL
reasoning has finished.

  - Usually, OWL reasoning does not identify something as incomplete,
because in an open world there always might be data somewhere that
completes it

  Therefore, I suggest to introduce the extensions as a separate
layer: They deserve their own name (maybe OWLCAGE?) and their own
syntax to avoid confusion.

  OWLCAGE reasoning builds on OWL reasoning and introduces new
assumptions. While OWL makes only conclusions that are still valid
when new data is added (open world assumption), OWLCAGE reasoning
makes conclusions valid only for a certain set of data (closed world
assumption). While OWL has no notion of completeness, OWLCAGE has a
well-defined notion that the present set of data is complete for a
particular purpose.

     Take care
     Oliver

On 10/10/07, Dan Corwin <d...@lexikos.com> wrote:

--
Oliver Ruebenacker, Post-Doc Researcher
Theoretical Biological Physics and Soft Statistical Mechanics
Cell Biology at UConn Health Center and Physics at Harvard
http://people.deas.harvard.edu/~oliver/