RE: AS&S and WG consensus (was Re: abstract syntax and RDFS)

> RDFS permits one to make statements about classes as instances.
> Why would you expect there to be a migration path from RDFS to OWL Lite?
>
> Is it that the migration you expect is only for certain subsets of RDFS,
> and you can't imagine those subsets not including rdfs:seeAlso?
>

It is clear that certain features of RDF and RDFS are not available in OWL
Lite.
The ones that I believe there is WG consensus about are:

- classes as instances, i.e. for three non-built-in URIrefs <a> <b> <c> you
may not have the following three triples:
  <a> rdf:type <b> .
  <b> rdf:type <c> .

- unrestricted use of blank nodes
   use of rdf:nodeID in an unrestricted fashion is not allowed in OWL Lite
e.g.
<a> <b> _:blank .
<a> <c> _:blank .
is not in OWL Lite.

- subproperties and subclasses of builtins.
 e.g.
  <a> rdfs:subPropertyOf rdfs:subPropertyOf .
  is not in OWL Lite

- properties taking both Literal and Object values


===

There are other features of RDF and RDFS that are excluded from the OWL Lite
discussed in AS&S that we have not discussed (or not discussed adequately)
and for which I do not yet see a WG consensus.

e.g.
- the prohibition of all use of xml:lang
- the prohibition of all use of rdf:parseType="Literal"
- the prohibition on using rdfs:label on both classes and instances
- the prohibition on using dc:creator on both classes and instances
- the prohibition of all use of rdfs:seeAlso, rdfs:isDefinedBy

Jeremy

Received on Monday, 27 January 2003 09:02:47 UTC