Re: Linkbases, Topic Maps, and RDF Knowledge Bases -- help me

[Uche Ogbuji <uche.ogbuji@fourthought.com>]

> <rdf:Description
>   about="http://spam.com#Malatesta"
>   type="http://art.org#Patron">
> [...]
>
> Didier replies
> I know you wont agree on this but what I would like to see is
> <topic xlink:type="extended" etc...>
> <resource xlink:type="locator" rdf:type="http://art.org#Patron"
> xlink:href="about="http://spam.com#Malatesta">
> .... property list for this frame .....
> </resource>
> </topic>
>
> In some ways an rdf element is a frame (like AI frames), if rdf:type is
> present it is no longer a free form frame but more a defined record, if the
> rdf:type is not present, it is a free form frame. This said, now the problem
> is for an RDF engine to recognize the this particular element is an rdf
> element.

Hmm.  I actually don't have a problem with your use of XLinks here, and
I might be misunderstanding you, but I think you might be getting a bit
hung up on the serialization.

But let me first of all see if I understand what you're saying.  Would you
find the following suitable?


<rdf:Description about="http://spam.com#Malatesta">
  <foo:name>Sigismundo Malatesta</foo:name>
</rdf:Description>

<rdf:Description about="http://spam.com#Malatesta">
  <rdf:type resource="http://art.org#Patron"/>
</rdf:Description>

or even

<art:Patron about="http://spam.com#Malatesta">
  <foo:name>Sigismundo Malatesta</foo:name>
</art:Patron>

both of which are equivalent to the original

<rdf:Description
  about="http://spam.com#Malatesta"
  type="http://art.org#Patron"
  foo:name="Sigismundo Malatesta"
/>

If one of these suit you, then I think you're drawing a false parallel
with frames.  Frames are more analogous to the resource in the resulting
abstract model.

> possible solutions:
> a) rdf recognizes xlink:href as equivalent to the "about" attribute, this
> would imply that an rdf frame is a link. hummm, not convinced that this is a
> good idea because it does not make sense for the other RDF usages.

I don't like this either.  XLink is too broadly used (and useful) for this
not to cause a potential explosion of noise in the model.

> b) to define a free form type like for example
> rdf:type="http://www.w3.org#frame". In this case we state that this element
> is a free form frame and not a record defined by a type. But if the rdf
> element is conformant to a certain structure and is therefore an instance of
> a certain class, then we would have instead
> rdf:type="http://art.org#Patron". I prefer this solution since it allows us
> to include a free form or defined collection of properties for a particular
> "locator".

Maybe I need to see more clearly the problem you're trying to solve.  The
*only* "built-in" thing an RDF type does not is provide for predicate
constraints.  One does not have to use these.  So you are free to view the
rdf:type as just one slot in the frame.

> I am not trying to convince you Oche, just showing that if elements could
> inherit the rdf frame behavior (defined record or free form frame) this
> would lead to a tremendous useful thing for knowledge management.

Please do try to convince me.  I hope I didn't make you think I was
implacable with the early business with "but".  I'm happy to be convinced,
abused, or disabused, as long as the agent in question is prepared to
accept the same treatment.  ;-)


-- 
Uche Ogbuji                               Principal Consultant
uche.ogbuji@fourthought.com               +1 303 583 9900 x 101
Fourthought, Inc.                         http://Fourthought.com
4735 East Walnut St, Ste. C, Boulder, CO 80301-2537, USA
Software-engineering, knowledge-management, XML, CORBA, Linux, Python