Peter noted that the recognition, subsumption, and consistency problems are computationally equivalent, therefore eliminating class expressions doesn't reduce the computational complexity. Sometimes the easiest implementation is to write the theorem prover. We may have to further reduce the meanings of features to get a complexity benefit.
We may also want to distinguish complexity for compiler writers vs. end users.
ACTION (Peter and Stefan): refine the options and proposal offline.
Ian noted that medical domain ontologies tend to record lots of machine-readable extra-logical information.
Peter expressed concern about the draft charter having 2 separate goals: ontology and rules. Ian reminded us of Ora Lassila's advice not to dilute the focus, and to go in with a strong proposal. The WebOnt rules effort may be staged, giving the Joint Committee and others time to experiment with rules and make proposals while the working group focuses initially on the ontology language.
$Id: 2001-06-12.html,v 1.3 2001/06/26 21:18:06 mdean Exp $