Re: OWL Reference Description for F2F 3

Here are my comments on the draft version.

peter

PS: I will not be at the face to face.  



		Comments on ``Reference Description''


The new reference description is a good start, but I think that it
needs lots of work before it can be even a working draft.  I enclose a
list of comments.  Many of these comments concern changes from
DAML+OIL to OWL or problems with the previous version of the document, not
problems introduced into the document by the current round of editing.


The title of the document does not correspond to its contents.  The
document is really an informal description of the OWL language, not a
reference description.  A better title is probably ``Language
Description.''

The list of other documents needs a lot of work, partly to reflect
changes in these documents.

References to the previous definition of RDF should all be changed to
the references to the RDF Core working drafts.

The description of the contents of the document should be changed to:

    This document informally specifies which collections of RDF
    triples constitute OWL knowledge bases and what the meaning of
    these knowledge bases is.

The paragraph on mixing OWL with RDF is misleading.  It needs to be
changed, but I'm not sure what to.  The sentence on the KIF
axiomatization should be removed, at least for now.

Does the header of an OWL ontology need to be at the beginning of an
OWL document?  If so, this is counter to the claim that all that
matters in an OWL KB is the triples.

All uses of ``definition'' should probably be removed from the document.
For the use of ``definition'' in the first Imports paragraph,
``information'' can be used instead.  The use of ``definition'' in the
first paragraph about Class elements should be changed to ``information
about''.  Other uses of ``definition'' may need different treatment.

Most uses of ``URI'' should be changed to ``URIref''.

The warnings about RDF Schema, should be removed, as RDF Schema has
changed.

Isn't disjointUnionOf gone?  If so, the paragraph on it should be
removed.

The caveat on several enumerations is misleading.  Having two enumerations
for a class just equates the two enumerations.  This may or may not be
inconsistent.  In any case, the result would be to make the KB
inconsistent, not the class.

Shouldn't rdfs:class be replaced with owl:class in boolean
combinations?

Datatype properties and object properties have similar, but not the
same syntax.  

The comment on multiple restrictions in a single restriction element
is wrong.  The effect is to equate the two restrictions, not form the
intersection.

The paragraph on restrictedBy should probably be removed.

Many of the notes could be inlined, particularly the one about boolean
combinations.

There should probably be a caveat about the use of owl:collection.
This is currently illegal, but may be accomodated by RDF Core.

Received on Tuesday, 18 June 2002 05:37:16 UTC