Re: [OEP] OWL restrictions

Guus Schreiber wrote:
>For a course that I'm teaching this month I have written a short note 
>about OWL restrictions, as this is something that students who start 
>working with OWL find difficult to understand. It is not a pattern note, 
>but maybe it could be useful as an add-on to the 101 paper of Natasha.
>
>You can find the note at:
>
>   http://www.cs.vu.nl/~guus/public/owl-restrictions/
>
>It still needs a careful read-through.

I haven't yet read the document all the way through, but I do have a 
few initial comments, most of which related to sections discussing UML 
(big surprise there ;).

1) You refer to UML as a data-modeling language in a couple of places in
the document.  Since even Class Diagrams in UML allow one to specify operations 
and visibility of attributes, object modeling is a more apt term.

2) In the text of bullet 3 in the section entitled, "Differences between
OWL properties and UML associations/attributes" you write, "UML associations 
and attributes need not to be unique;".  It would be more precise to say
that "Names of UML associations and attributes need not be unique"  These 
are actually distinct elements in a UML model, its just that the naming
rules don't require distinct names for them when the context is sufficient
to distinguish them.  Simply put, the naming scopes are very different in
rdf/owl vs UML.  The name scope for an attribute is the class that contains
it and the name scope an association is the combination of the package
containing it and the classes it associates.

Also when you contrast OWL/RDF properties as "first-class citizens", you
say "they have an identify, i.e. a URI such as http://www.example.org/wine#color)."
I suggest changing this to read "they have a global identity."  

3) In the "UML Note" you discuss cardinality constraints (UML:Multiplicity)
on attributes. You are quite right in noting that attributes can be multivalued,
but the syntax in the example is slightly off.  The syntax for a multiplicity
allowing any number of values would be [*] or [0..*] where the first form is just
an abbreviation of the second form.  See section 3.44 of the UML 1.5 specification
[1] for valid multiplicities.

*****

Kudos for the guideline in this section.  People should be designing web ontologies
for distributed (re)use and not putting unnecessary restrictions on properties.

*****

As an aside.  In contrast to your experience, I do not find N3 easier to read then
rdf/xml.  However, I can imagine it is a lot less tedious to write.

BTW - Was there an editing error that accidently appended a definition for 
Committee to the RedThing example of a necessary and sufficient condition? 

[1] http://www.omg.org/docs/formal/03-03-01

Received on Friday, 3 June 2005 20:43:55 UTC