Re: Weak entity classes and the part-of association

On Fri, 2003-11-21 at 06:01, Saied Tazari wrote:
> As discussed in [1]:
> 
> "
> Embedded resources result from either the part-of association – one of 
> the fundamental concepts in object-oriented modeling – or the theorem of 
> weak entity classes in database management. In UML class diagrams, for 
> example, associations having a diamond on one side indicate that the 
> class on that side represents composite objects having instances of the 
> class on the other side as their parts. They go even further and say 
> that if the diamond is darkened, then the instances of the class on the 
> other side may never exist independently from instances of the class on 
> the side of the diamond – quite similar to the concept of weak entity 
> classes in database management.
> However, there is currently no standard way to specify part-of 
> associations as such in RDF schema or OWL. CC/PP [8], as an RDF-based 
> approach for articulation and exchange of contextual knowledge in 
> profiles, has proposed the concept of components that can be seen as a 
> solution for this problem. It was not consistent enough, though. It 
> seems that the understanding of profiles in CC/PP is something like the 
> old ".ini"-files in Windows(TM); i.e. all attributes must appear in some 
> component and no attributes within components may have another component 
> as value. None of these restrictions matched our requirements. This 
> approach ignores the evolution of such config-files into tree structures 
> like Windows(TM) registry.
> "
> 
> To improve OWL in this regard, I suggest to add the following concepts:
> 
> hasComponent|hasPart        rdf:type        TransitiveProperty
> componentOf|partOf          rdf:type        TransitiveProperty
> componentOf|partOf          inverseOf       hasComponent|hasPart
> WeakClass|WeakEntity        rdfs:subClassOf Class
> hasContainerClass|dependsOn rdf:type        TransitiveProperty
> hasContainerClass|dependsOn rdfs:domain     WeakClass|WeakEntity
> hasContainerClass|dependsOn rdfs:range      Class
> 
> Other statements may also be added, e.g. about the cardinality of 
> 'hasContainerClass' for instances of 'WeakClass' (it must be greater 
> than 0).
> 
> Of course, 'hasComponent' and 'componentOf' should be used in the 
> descriptions of instances only when no additional semantics of the 
> corresponding relation is known.  Usually, however, one should introduce 
> appropriate sub-properties of them and use the sub-properties in order 
> to imply the additional semantics.

This is a very interesting suggestion, but I wonder if this
version of the OWL specification is the best place for it.

As we said in the status of the 31 March requirements draft...

"Requests for significant changes to the requirements are not
anticipated and will be evaluated in the context of the scope and
schedule of the Web Ontology Working Group charter and other plans for
the W3C Semantic Web Activity"
 -- http://www.w3.org/TR/2003/WD-webont-req-20030331/

As work on this version of OWL winds down, we have started
talking about best practices for using OWL and such.
Coincidentally, part/whole relationships is one of the first
things we've started writing about in this context:
  http://esw.w3.org/topic/PartWhole

See also discussion of a charter for a possible Best Practices
working group:

  SW Best Practices WG - draft charter for discussion 
  Guus Schreiber (Monday, 10 November)
  http://lists.w3.org/Archives/Public/www-rdf-interest/2003Nov/0055.html


Does it seem satisfactory to take up your suggestion in
a later stage of work? Note the OWL drafts have been
out since March 2002 and we were really hoping
to get all comments on the OWL specs in by 20 September 2003.


> Regards,
> 
> -- Saied
> 
> [1] Tazari, M.R.  A Context-Oriented RDF Database.  Semantic Web
>      and Databases, Berlin (Germany), September 2003.
>      http://www.zgdv.de/~stazari/publications/SWDB03.pdf
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Wednesday, 26 November 2003 17:27:09 UTC