[no subject]

Now of course this is just using the current registry
classification system to categorize these - but gives
you a useful start - and a means to use the power
of CAM in a UDEF context.  I'm postioning this
as a useful first base to get to - so that we can
then understand what more is needed after that.

Thanks, DW.

----- Original Message ----- 
From: <ewallace@cme.nist.gov>
To: <regrep-semantic@lists.oasis-open.org>
Sent: Friday, February 06, 2004 2:41 PM
Subject: [regrep-semantic] re: Will a Universal Data Element Framework
[UDEF] Class of Applications fit within the ebXML Semantic Registry ?


> Zachary Alexander wrote:
>
> >Will a UDEF (Universal Data Element Framework) Class of application fit
> >within the ebXML Semantic Registry model? How would a Semantic Aware
> >ebXML Registry support UDEF? The UDEF is an international,
> >cross-industry standards effort that is developing Object and Property
> >word trees that can be combined to construct semantic identifiers.
> >
> >[1] http://www.udef.org
>
> I expected UDEF to come up in this group sooner or later.  IMHO it is
> out of scope for this group because the trees it uses for encoding
> things are not quite taxonomies.  (They are not taxonomies because the
> parent-child relationships for nodes in the tree are not always
> subsumption and because concepts are actually denoted by the paths
> from these nodes to the root rather than the node itself). This puts
> UDEF to the left of taxonomy on Leo's chart, out of the area that
> could be described as Ontology or Semantic Model.
>
> The UDEF Object tree also provides a good example of the problems of
> trying to encode a large array of divergent concepts into a simple
> single tree.  Some notable results are: different interpretation of
> parent-child relationships even at a single parent; and multiple
> occurences of the same word at different levels and in different
> branches.  This makes identification of a concept difficult and keyword
> searches for a concept not very interesting.  The former result points
> the need for multiple kinds of relationships in conceptual models,
> while the latter result points to the need to support lattices of these
> relationships.
>
> -Evan
>
>