documents vs. sublanguages

I'm getting increasingly concerned about the notion of
associating OWL sublanguages (Lite/DL/Full) with documents
(as opposed to tools and other processors of those
documents).  We probably made the right decision to postpone
sublanguage-tagging of ontologies.

I think we want to encourage ontologists to express
"everything" they know about their application domain,
without major concern for the language level(s) of the
consumer(s).  If some tools ignore some of this information,
so be it.

The alternatives are pretty bleak:

  never express anything beyond the lowest common
  denominator (OWL Lite or RDFS)

  maintain separate versions of an ontology for different
  language levels - a configuration management nightmare

  provide some sort of mechanism like "#ifdef owldl ..."

This isn't to say that there isn't some utility for a tool
that can identify ontologies/content for which, e.g., an OWL
Full or OWL DL reasoner would offer value beyond an OWL Lite
reasoner.

Given this view, our tests should probably each identify the
expected results for each of our sublanguages for the given
test input (at least in cases where the results are expected
to be different).

	Mike

Received on Friday, 7 February 2003 13:39:58 UTC