W3C home > Mailing lists > Public > public-webont-comments@w3.org > April 2008

Comment on OWL 1.1

From: David Poole <poole@cs.ubc.ca>
Date: Thu, 03 Apr 2008 23:05:03 -0700
Message-ID: <47F5C50F.9020600@cs.ubc.ca>
To: public-webont-comments@w3.org

Hi!

Here are some random comments on the OWL 1.1 proposal. I am writing
this with two roles: we are writing ontologies in OWL and we are
writing an introductory AI textbook and want to include OWL. My main
comments are with the second role. We want to keep it simple, but not
oversimplify; which, as you know, is difficult.

1. Getting rid of "onProperty" is good (this is the main reason I
    would rather use 1.1 than 1.0 in our book).

2. There is too much proliferation of terms. It is *much* better to
    have just ObjectProperty and DatatypeProperty and use these
    conjoined with the other class constructors or properties, rather
    than duplicating virtually every class constructor. This is one
    area where 1.1 is worse than than 1.0 for no apparent gain.

    The 1.1 constructs are less expressive and more verbose than 1.0
    constructs with no particular advantage. For example, I may want to
    write an axiom that only depends on whether the property is
    functional, and not care if it is an object property or a
    functional property. In the 1.1 constructs it is a real pain to
    do this.

3. You need to make up your mind what the elements of a class
    are. They either should be "individuals" or "objects" (I don't care
    which, but please be consistent), then use this terminology
    consistently in OWL and in the documentation.
   - If you want to use "object" then have "ObjectProperty" is a
      property whose value is an object. But then you should have
      "sameObject" not "sameIndividual".
   - If you want to use "individual" then you should have
    "IndividualProperty" is a property whose value is an individual.

    At the moment you state in the http://www.w3.org/TR/owl11-syntax/
    that "OWL 1.1 objects (ontologies, axioms, etc.)" but then
    ObjectProperty is a property whose value is an individual. The
    terminology should be consistent between the syntax and the
    semantics documents and the language itself.

Thanks for the work you have put into this!

David

-- 
David Poole,                      poole@cs.ubc.ca
Department of Computer Science,   http://www.cs.ubc.ca/spider/poole
University of British Columbia,   Office: +1 (604) 822-6254
Vancouver, B.C., Canada V6T 1Z4   Fax:    +1 (604) 822-5485
Received on Friday, 4 April 2008 09:53:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:43:30 GMT