Re: Layering on RDFS

Just for clarity the message was originally sent to my HP colleagues and so 
reflects a greater attempt at clarity than I really percieve within the WG. 
The paragraph you quote is my extrapolation from where the WG is currently 
at, rather than my sense of where the WG is.


I think the complicated network within large owl, which you speak of, will 
be largely unavoidable; because we don't have an implementation strategy 
that permits all the features that everyone wants.

As I  indicated in the message, having large owl is a step forward because 
it at least gives a common framework in which all these features are specified.

Jeremy


Masahiro Hori wrote:

> On Fri, 2002-10-16 at 17:02, Jeremy Carroll wrote:
> 
>>In practice that means that we expect the full functionality of Fast OWL
>>to be implemented, whereas some aspects of Large OWL will not be, and we
>>will not define a minimal requirement for Large OWL implementations over
>>and above Fast OWL.
>>
> 
> If a minimal requirement for Large OWL is not defined, we can have
> an implementation of Large OWL that is solely given with the Classes
> as instances feature.  I am cuorious if this implementation is usable
> or not for those who want to have the Classes as instances feature in
> the OWL Lite.
> 
> 
> Another point is if a minimal requirement for Large OWL is not defined,
> the Large OWL is not a language but a class of ontology languages.
> This means we might have a complecated "network" of languages within
> the Large OWL(s).  However, I still strongly support to have the inclusion
> ordering: OWL Lite < OWL/FOL < Large OWL, because it makes easier
> for ontology users to decide which OWL should be used for individual
> situations.
> 
> 
> -Masahiro Hori
> 
> Masahiro Hori, Ph.D.
> Group Leader, Programming Models & Tools,
> IBM Tokyo Research Laboratory
> Tel: +81-46-215-4667 / Fax: +81-46-274-4282
> Email: horim@jp.ibm.com
> 
> 
>

Received on Friday, 18 October 2002 16:45:02 UTC