Re: XML Schema for OWL

From: "Raphael Volz" <rvo@aifb.uni-karlsruhe.de>
Subject: XML Schema for OWL
Date: Wed, 19 Jun 2002 17:08:19 +0200

> Hi Peter -
> 
> I finally had time to have a look to your XML Schema.

[...]

> Regarding the definition of OntologyType:
> -----------------------------------------
> Why do you use the plural for the elements "EquivalentClasses",
> "DisjointClasses","EquivalentProperties","DifferentIndividuals" ?

Because they all take multiple classes (or properties or individuals) and
do not equate their arguments.

> With respect to "EquivalentClasses" I do not understand how your example can
> validate against the schema. In the example you use the singular form
> "EquivalentClass"  to declare "Student".

The example corresponds to an older version of the schema.  Per a
suggestion EquivalentClass changed to Class with an owl:complete="true".
The example has been changed to the new way.

> Is the ability to specify SubClassOf and SubPropertyOf separately from the
> Class and Property definition directly
> under the ontology definition intended ?

Yes.

> Regarding ClassAxiom:
> ---------------------
> I can't understand the last attribute in the definition of class axioms:
> 
>  <xsd:complexType name="ClassAxiom">    <!-- RDF description ...
>   <xsd:attribute complete="complete" type="xsd:boolean" use="required" />
>  </xsd:complexType>
> 
> There is no attribute "complete" in XML Schema for xsd:attribute ( see
> http://www.w3.org/TR/2001/REC-xmlschema-1-20010502/#element-attribute )

Sorry, this should be 

>   <xsd:attribute name="complete" type="xsd:boolean" use="required" />

See above for what this attribute is for.


> Regarding the definition of "SubClassOfType":
> ---------------------------------------------
> You can omit the minoccurs and maxoccurs for sub and super elements since
> they take the default values anyways.
> 
> The same holds for SubPropertyOfType.
> The same holds for DataType / OnfeOf in DataRange.
> The same holds for Class etc. in DescriptionType.
> The same holds for SingleDescription.
> The same holds for RestrictionType.

I put them in to make them explicit.  


> Within DataRestrictionType and IndividualRestrictionType:
> ---------------------------------------------------------
> is the ability to specify arbitrary numbers of allValuesFrom and
> someValueFrom (at the same time) really intended ? To me this does not make
> sense and should be appropriately restricted.

Yes.  It is useful to put in several of these at one time.  (It is, of
course, more useful for someValueFrom than for allValuesFrom.)

> General Remarks:
> ----------------
> I'ld suggest that a general pattern for the definition
> of multiple occurences is used.
> 
> Sometimes you establish cardinalities with the sequence
> (for example at the definition of OntologyType) and sometimes
> with the element itself (for example at the definition of Enumeration).

Well there is a difference between the two.  Admittedly, the OntologyType
form could be used in Enumeration.

> Also I'ld suggest to have a general pattern for naming, which
> is partially there, e.g.:
> 
> Ontology -> OntologyType
> Include -> IncludeType
> SubClassOf -> SubClassOfType
> 
> on the other hand you use:
> Class -> ClassAxiom
> EnumeratedClass -> Enumeration
> ...

Yes, this would be a good idea.  To be done.

> Third the requirement of having the readability of the
> XML through RDF leads to quite many spare nestings which
> will lead to even more anonymous triples in the very end.

Well, it does lead to some extra nesting.  Not too much, however.

> Best regards,
> 
> Raphael
> 
> PS: I'll try to come up with an alternative concerning
> the last problem but this may take a while to get done.

You might want to look at 

	http://www-db.research.bell-labs.com/user/pfps/owl/schema-2.xsd
	http://www-db.research.bell-labs.com/user/pfps/owl/schema-2-example.xsd

Note that there are errors remaining there.


>  Raphael Volz             Tel:   49 721 608-7363

peter

Received on Wednesday, 19 June 2002 12:32:00 UTC