W3C home > Mailing lists > Public > public-owl-dev@w3.org > January to March 2007

Structural Specification / XML Schema comments

From: Thorsten Liebig <thorsten.liebig@uni-ulm.de>
Date: Mon, 22 Jan 2007 16:45:22 +0100
Message-ID: <45B4DC12.8030406@uni-ulm.de>
To: public-owl-dev@w3.org

Hi all,

the following contains some remarks to the latest (11/17/06) "Structural
Specification and Functional-Style Syntax" document as well as to the
corresponding XML Schema of OWL 1.1:

o The recently announced "InverseObjectProperties" object property axiom
  is missing in Figure 12.

o Since "SameIndividual" is applied to a set of individuals it should be
  renamed to "SameIndividuals" in Figure 15 as well as in the
  corresponding XML schema.

o The DisjointUnion axiom defines a class as a union of other classes.
  The XML schema, however, forces to give at least 2 classes to define.
  As far as I see line 442 of the schema should be:
  <xsd:group ref="owl:Description"/>

o Suggestion to Figure 12: A "subObjectProperties" may contain an object
  property chain as first argument. This is reflected with help of a
  <<list>> association requiring at least 1 object property of type
  ObjectPropertyExpression. This does not explicitly represent the
  notion of object property chains and may therefore easily be overseen
  or misunderstood. Why not introducing "ObjectPropertyChain" as
  specialization of "ObjectPropertyExpression" for the sake of clarity
  of object property axioms.

o The second sentence of section 8 (Declarations and Structural Consistency)
  seems to be somehow misleading. It states that declarations "have no
  semantic effect in an ontology". However, when providing a declaration for
  a class, say "A", which is not referenced elsewhere, this class will be
  returned when asking for the child classes of owl:Thing. It seems to me,
  that a declaration is some special SubClassOf axiom, namely
  SubClassOf(A owl:Thing), with some extra annotation feature (to annotate
  entities such as classes).
  As already discussed within the DIG 2.0 working group, I still don't
  see much benefit in having declarations. Typographical errors in
  descriptors should be handled by authoring tools and not by reasoners in
  my opinion.


Thorsten Liebig                        Institute for AI
Tel.: +49 731 502 4207      Faculty of Computer Science
Fax.: +49 731 502 4119                University of Ulm
                                    89069, Ulm, Germany
Received on Monday, 22 January 2007 15:45:43 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:58:14 UTC