Re: [regrep-semantic] Does Taxonomy = Semantic Content Management???

[ewallace@cme.nist.gov]

"Zachary Alexander" <zack2003@ebtdesign.com> wrote:

>><David RR Webber>Now here's the other bit of good news - the ebXML
>>Registry with its classification system and RIM already has a giant
>>chunk of what we need to get a Version 1.0 system up and functioning.  I
>>think that should also be our mantra here too - look for lowhanging
>>fruit - that will enable significant progress, and worry about the
>>really clever tough stuff after we've got some solid metrics to base our
>>understanding on.</David RR Webber>
>
>The ebXML Registry already supports taxonomies.  Support for taxonomies
>equate too roughly to OWL Lite.  Does taxonomy support (i.e., roughly
>OWL Lite) constitute Semantic Content Management Support? Does that set
>the bar high enough to entice members of the Semantic Web Community to
>choose the ebXML Registry over some other "knowledge repository" that
>supports OWL DL or OWL Full?  
>

OWL Lite does indeed support defining taxonomies, but it is not limited to
just that.  A quick look at the OWL Lite Synopsis in the "OWL Web Language
Overview" document [1] shows that it also has vocabulary supporting 
specification of relationships and attributes (rdf:Property), specification 
of properties with particular mathematical characteristics (transitive, 
symmetric, functional, inversefunctional), specification of limited 
cardinality restrictions, and limited quantification.  With this one can
define ontologies and not merely taxonomies.  In addition, OWL Lite 
provides vocabulary elements for linking ontologies (equivalentclass, 
equivalentproperty, etc).  

I am not aware of what ebXML Registry already supports, but if it is
just subsumption trees then OWL Lite will provide considerably more
expressive power.  Having said that, I wouldn't want to impose a restriction
against OWL DL or OWL Full as valid semantic content for this repository.

> 
>Note: OWL Lite, OWL DL and OWL Full are just examples that show higher
>levels of semantic/knowledge expressivity.

The criteria for the restrictions on OWL Lite and OWL DL were based
on ease of implementation of reasoners for those dialects.  That's why
they really don't fit nicely into anyone's broad classes of logical
models.  I agree that these are somewhat arbitrary restrictions of OWL
Full.  There certainly could be reasons for using a different restriction,
but it is still useful to have names for a couple points on the spectrum that
we can all refer to.

-Evan




[1] http://www.w3.org/TR/owl-features/