Re: DC with RDF Modeling Questions

kurt.godden@gm.com wrote:
> I'm developing an RDF application, but I have some questions that I have
> not found in the W3C documentation, in case any kind soul feels like
> helping me.
> 
> It is clear that I should re-use existing ontologies whenever possible.
> Thus, I am using some of the Dublin Core elements/refinements, e.g. I'm
> using dc:title to hold the primary info of a class called "EngineeringName"
> that I'm using to encode a list of existing corporate terms.

> Q2:  Is the use of QNames ok in the value of an rdf:ID attribute, as with
> the following 'dc:title', where I also have the appropriate namespace
> assignment to the prefix?
> 

No, use the full URI for the property.  You can define entities in a DTD 
to get an abbreviation:

	rdf:ID='&dc;title'

> <rdf:Property rdf:ID="dc:title"/>          <--- Is 'dc:title' OK to use
> this way????  Should I use an entity instead of the prefix????

Yes, use an entity as as above.  No QNames in attribute values, please.

> Separate topic:
> 
> Each engineering object, e.g. a body control module, gets realized in
> physical form as parts shipped from suppliers.  For a given engineering
> object, there may be more than one corresponding part number (e.g. PN 1234
> from supplier A and PN5678 from supplier B may both be functionally
> equivalent body control modules).  I want to have a pointer from the
> resource representing the engineering object to a bag of such part numbers
> that represent the physical realizations of that object.
> 
> Q3:  It seems to me that a part number is a version/edition/adaptation of
> an engineering name, and thus it is reasonable to model the relationship
> with dc:hasVersion,

I think that's stretching the meaning of version, but to some extent 
that depends on the point of view of the user/definer.  I would suggest 
that equivalentClass (an OWL concept, not a dc one) would be more 
suitable (though maybe not quite right either, depending on your 
viewpoint, since both equivalent classes are supposed to have the same 
class extensions).    Maybe a union of two classes, one for each vendor, 
would best express your intent.  After all, you probably have your own 
part number that can be realized with either of the two, right?  So 
define a class for your part number that is the union.

Cheers,

Tom P
-- 
Thomas B. Passin
Explorer's Guide to the Semantic Web (Manning Books)
http://www.manning.com/catalog/view.php?book=passin

Received on Tuesday, 20 July 2004 17:54:04 UTC