W3C home > Mailing lists > Public > public-webont-comments@w3.org > January 2003

Re: Comments on Owl Reference Ed's draft

From: Guus Schreiber <schreiber@swi.psy.uva.nl>
Date: Fri, 31 Jan 2003 22:38:35 +0100
Message-ID: <3E3AECDB.2040108@swi.psy.uva.nl>
To: Brian McBride <bwm@hplb.hpl.hp.com>
CC: public-webont-comments@w3.org


Thanks for your elaborate review. See for comments inline.
The review was very helpful and has contributed to substantial
revision of the document.

The revised version will become available within a couple of days as an
editor' draft. We will let you know when that happens.


Brian McBride wrote:
 > I've been reading through the editor's version of the owl guide at:
 >    http://www.daml.org/2002/06/webont/owl-ref-proposed
 > version
 >   $Id: owl-ref-proposed.html,v 1.112 2003/01/03 04:35:38 mdean Exp $
 > The editors may find the following comments useful.  These comments are
 > from me alone in my capacity as an interested individual.  I have marked
 > those I feel deserve a response, though they are largely editorial in
 > nature, with SUBSTANTIVE or TECHNICAL.
 > I realise that I am not referring to a published WD and am commenting on
 > work in progress.  However, I hope the editors may find these useful 
in the
 > final stages of polishing this document.
 > Some of these comments are duplicates of ones I have made before
 > but are not listed in the editor's list of comments being addressed, so
 > perhaps bear repeating, lest they were missed.
 >    http://lists.w3.org/Archives/Public/www-webont-wg/2003Jan/0071.html
 > 1.
 > The abstract refers to the RDF/XML syntax doc.  The concepts doc may be
 > more appropriate.

Refs from abstract have been removed. RDF concepts document is cited
at the first place where RDF is mentioned in the running text.

 > 2.
 > [[    * OWL Lite is detailed in the Feature Synopsis for OWL Lite and 
 > [OWL Features].]]
 > "Detail" is not a verb.  What does this sentence mean?

The phrase dpes not occurr anymore.

 > [[An OWL knowledge base is a collection of RDF triples as defined in the
 > RDF/XML Syntax Specification (Revised) [RDF/XML Syntax].]]
 > Again I would have expected the concepts doc to be the appropriate
 > reference here.

This part has been signifanctly changed.

 > 4.
 > [[This document specifies which collections of RDF triples constitute 
 > OWL vocabulary and what the prescribed meaning of such triples is.]]
 > The owl vocabulary is a collection of triples?  That seems like a type
 > mismatch.  I'd expect a vocabulary to be a collection of names, in this
 > case RDF URIREF's.

This was an error. It now reads: "An OWL ontology is in fact a set of

 > [[OWL only provides a semantic interpretation for those parts of an RDF
 > graph that instantiate the schema defined in 
 > What does it mean for "part of an RDF graph" to "instantiate" a 
schema?  Is
 > the term "instantiate" defined anywhere?  RDFS has no such notion.
 > I'm finding it a bit bizarre that RDF Schema is being used to define the
 > syntax of a language.

This sentence has been removed and the section has been rewritten (see
Sec. 1.4).

 > owl:backwardCompatibleWith
 > What does this actually mean?
 > Given:
 >    o1  owl:backwardCompatibleWith o2 .
 > does that entail
 >    o1#foo owl:sameAs o2#foo .
 > for all foo defined in o2.   Does that include foo's that are in the o2
 > 'namespace' but are not actually defined in the ontology which can be
 > retrieved from o2?  What is a sufficient condition to allow stating that
 > one ontology is backward compatible with another.  How are imports 
 > by owl:backwardCompatibleWith?  If an ontology drops an import can it 
 > be backward compatible with its previous version.  If it changes an 
 > to one that is backward compatible, is it still backward compatible?
 > I think the meaning of this needs spelling out more clearly.

Currently, all version statements have no logicakl meaning except for
the RDF meaning. This is now explcitly mentioned in the document.

 > [[OWL divides the universe into two disjoint parts. One part consists of
 > the values that belong to XML Schema datatypes. This part is called the
 > datatype domain.]]
 > RDF allows for datatypes that are not XML Schema datatypes, but 
conform to
 > the XML Schema datatype model.  The text here rules out such 
datatypes.  Is
 > that intended?
 > Later the doc states:
 > [[OWL also allows the use of XML Schema datatypes to describe (or 
 > part of the datatype domain.]]
 > I think this is contradicting the previous statement, but the language
 > isn't very clear to me.  Is this a syntactic statement - that xml schema
 > datatype declarations are allowed as part of an owl ontology?

This has been completely rewritten. These sentences do not occur
anymore. See Sec. 7 on Datatypes.

 > [[RDFS and OWL allow cycles of subclass relationships as a means to 
 > equality between classes.]]
 > Can we be clearer about what is meant by equality here?  I think you 
 > that a cycle of subclass relationships states that the classes in the 
 > have the same membership, but not that they are identical.

This sentence now does not mention class equality, but just states
that such cyclic staements have the effect of enforcing the same class
extension for both classes.

 > [[ Each owl:sameClassAs element asserts that C is equivalent to the
 > class-expression in the element (ie. C and all the class-expression must
 > have the same instances);]]
 > Is there a first class notion of class equivalence here?  If so it may
 > deserve more weight than a parenthetical comment.
 > Given the semantics, the name 'sameClassAs' is misleading.

A WG decision has been made to rename "sameClassAs" to
"equivalentClass". All documents will be updated to reflect this. The
documents will make explcit statements about
the difference between
class equality (the same concept, the same intensional meaning) and
class equivalence (the same values, in this case the same class

 > I note that owl:sameClassAs is not defined as a subproperty of
 > owl:sameAs.  Should it be?
 > I'm coming back to this comment after reading further on.  It maybe 
that it
 > should not be, but its not really clear to me.  I'm wondering what 
 > the semantics of owl:sameAs are.  In owl full, given X to be a class, a
 > property and an individual would:
 >    X owl:sameAs Y .
 > entail
 >    X owl:sameClassAs      Y .
 >    X owl:samePropertyAs   Y .
 >    X owl:sameIndividualAs Y .
 > If that is the semantics then owl:sameClassAs should not be a 
 > of owl:sameAs, since that would prohibit one saying that two classes 
 > the same members, but are not the same individuals.
 > But, thinking about it, that can't be the semantics can it, because 
if it
 > were, then one would not be able to use owl:sameAs in place of
 > owl:sameClassAs as suggested in the doc, as they don't mean the same 
 > in full owl.  I'm confused.
 > It would be useful to have the definition of the meaning of 
owl:sameAs all
 > in one place.

All this stuff has gone:

"sameAs" is a synonym of sameIndividualAs (this was a decision of a
while ago, but was never properly reflected in the document).
This is the "real" eqiality relation. sameAs has individuals as its
domain and range, so class equality can can only be exoressed in OWL
Full (which allows treating classes as individuals).

W.r.t. "samePropertyAs": this has (in a similar vein as sameClassAs)
been renamed to equivalentProperty, with a similar meaning (same
property extension / set of tuples/pairs/instances).

 > 11.
 > [[#  The class C must be equivalent to the class defined by each of the
 > boolean class expression,
 > and ..]]
 > I assume that the equivalent here is the one defined in parenthesis
 > above.  An internal link would be useful where such first class 
notions are
 > used.

Wording has been changed.

 > [[    * a boolean combination of class expressions, enclosed in
 > <rdfs:Class>...</rdfs:Class>  tags
 > ]]
 > Was <owl:Class>...</owl:Class> meant here?  If not, some comment on 
why not
 > might be appropriate.

Has been changed to owl:Class

 > [[Each class expression either refers to a named class, namely the class
 > that is identified by the URI, or implicitly defines an anonymous class,
 > respectively the class that contains exactly the enumerated elements, 
 > Should that be "...respectively a class that contains exactly...".  The
 > current text suggests that there is only one such class.  A similar 
 > applies to each of the following clauses.

In many places in the document this change from "the class" to "a
class" has been made.

 > [[Two class names are already predefined, namely the classes 
owl:Thing and
 > owl:Nothing. Every object is a member of owl:Thing, and no object is a
 > member of owl:Nothing. Consequently, every class is a subclass of 
 > and owl:Nothing is a subclass of every class.]]
 > I take these are the definitions of owl:Thing and owl:Nothing.  The
 > defining text
 >    [[Every object is a member of owl:Thing]]
 > does not rule out things other than objects being members of 
owl:Thing and
 > it should.  Similarly for owl:Nothing.

Wording has been changed for both to rule out other types of mebers.

 > [[A property restriction is a special kind of class expression. It
 > implicitly defines an anonymous class, namely the class of all 
objects that
 > satisfy the restriction. ]]
 > Similar to above, is it "the class of all objects..." or "a class ..."

See above.

 > [[An owl:Restriction element contains an owl:onProperty element, which
 > refers to a property name (a URI) ...]
 > should that be "(an RDF URI Reference)" with a link to the appropriate
 > definition or are OWL property names restricted to being URI's?

In a number of places URI has been changed to URT reference.

 > [[ An owl:allValuesFrom element defines the class of all objects for 
 > the values of property P...]]
 > Similar to above, should that be "a class of ..."

Yes, see above.

 > 18. typo?
 > [[an owl:someValuesFrom  element (which contains a class expression 
or a
 > datatype references).]]
 > Should that be "datatype reference)."?

It is now called a data range.

 > 19. typo?
 > [[observe that the owl:allValuesFrom  restriction demands that all 
 > of P belong to class P]]
 > Did you mean class *P*.

No. Changed.

 > [[an owl:intersectionOf element, containing a list of class expressions.
 > This defines the class that consists of exactly all the objects that are
 > common to all class expressions from the list. ]]
 > As above should it be "a class"?  Similarly for owl:unionOf and
 > owl:complimentOf.

Chaged , as above.

 > [[An rdf:Property element refers to a property name (a URI) (to which we
 > will refer as P)]]
 > Should that be an RDF URI Reference rather than a URI?  Is it the 
 > that is referred to as P or the property element.  None of these I 
 > as the doc later discusses subproperties of P, which makes P a property.
 > There is a similar issue with the earlier wording:

changed, see above.

 > [[A class element, owl:Class, contains (part of) the definition of an
 > object class. A class element refers to a class name (a URI) (we will 
 > to this class as C) and contains...]]
 > What class?

This part has been completely rewritten.

 > 22.
 > It seems a bit strange to be defining the semantics of 
 > rdfs:domain and rdfs:range here with no reference to the RDF schema
 > definitions.

This part now points to the relevant RDF documents, and discusses OWL
specifc aspects of using these (e.g. using class descriptions for domain
or ranges).

 > 23.
 > [[zero or more owl:samePropertyAs elements (each containing a 
property name).
 > Each owl:samePropertyAs element asserts that P is equivalent to the 
 > property (i.e. they must have the same instances),]]
 > If this is a first class notion of property equivalence, then it 
 > greater status than a parenthetical comment.
 > I also suggest that it, like owl:sameClassAs, is misnamed since I assume
 > (it not being clear from this document so far) that properties are 
 > intensionally.  owl:equivalentTo perhaps.

See earlier remark, now renamed to equivalentProperty

 > Should owl:samePropertyAs be defined as a subproperty of owl:sameAs?
 > Similar comment to 10.

No, see earlier comments.

 > We have part of the definition of owl:sameAs here.  But if we follow the
 > link to to the definition of owl:sameAs
 >    http://www.daml.org/2002/06/webont/owl-ref-proposed#sameAs-def
 > we find ourselves in the section on Class elements, which clearly 
does not
 > contain a complete definition of owl:sameAs. That doesn't seem like a 
 > thing.  I wonder if there is any more to owl:sameAs we haven't heard of
 > yet, e.g. can it be applied to an individual?  Checking
 >    http://www.daml.org/2002/06/webont/owl
 > it has no range constraint so presumably it can be.  But where is its
 > meaning defined?  But we also notice ...

In the revised owl.owl it says:

owl:sameAs equivalentProperty sameIndividualAs .

 > in
 >    http://www.daml.org/2002/06/webont/owl
 > we find:
 > [[<Property rdf:ID="sameAs"> <!-- equals? equiv? renames? -->
 >    <rdfs:label>sameAs</rdfs:label>
 >    <comment>
 >      for sameAs(X, Y), read X is an equivalent term to Y.
 >    </comment>
 > </Property>]]
 > I'm confused by the comment.  I'm not sure what a 'term' is here.  Is it
 > needed?  Does not "read X is equivalent to Y" do just as well.  Or is 
 > some meaning to "term".

Comments in owl.owl were out of date and have for the moment been removed.
We'e considering putting them back, but in the form of isDefined
links (this requires some semanitc problems to be solved.).

 > [[ Each owl:inverseOf element asserts that P is the inverse relation 
of the
 > named property.]]
 > Similar to the "the class" comments above, should that be "that P is an
 > inverse of the named property"?

Changed as suggested.

 > 28.
 > [[To state that objects are the same, an owl:sameIndividualAs element is
 > used. (Note that owl:sameAs can be also used here, but 
 > is preferred.)]]
 > More of the definition of owl:sameAs.   Should be collected into one 

Done in new version.

 > [[See issue #I5.18-Unique-Names-Assumption-Support-in-OWL. The 
situation is
 > different for datatype values, where XML Schema Datatype identity is 
used. ]]
 > You need to be clear exactly what identity is meant here.  There are 
 > with schema datatypes where the same value from the value space of
 > different datatypes is not considered by xml schema datatypes to be 
 > [[By deprecating a term, it means that the term should not be used in 
 > documents that commit to the ontology.]]
 > I'm not sure what is meant by term here.  Does it mean the name, or the
 > thing denoted by the name.  Given:
 >    C rdf:type owl:deprecatedClass .
 > does that mean that the name 'C' is deprecated or the class C.  Test 
 > does:
 >    C rdf:type              owl:deprecatedClass .
 >    D owl:sameIndividualAs  C .
 > entail
 >    D rdf:type              owl:deprecatedClass .
 > I'm assuming the answer must be yes, but in which case the reference to
 > "term" isn't appropriate.

I will be asking Jeff Heflin to answer this final point.

 > Done for now.  I hope that this is helpful.
 > Brian

Thanks, Guus
Received on Thursday, 30 January 2003 16:37:28 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:09:28 UTC