- From: Frank van Harmelen <Frank.van.Harmelen@cs.vu.nl>
- Date: Wed, 19 Dec 2001 00:02:30 +0100
- To: Enrico Motta <e.motta@open.ac.uk>
- CC: pfps@research.bell-labs.com, phayes@ai.uwf.edu, jjc@hplb.hpl.hp.com, horrocks@cs.man.ac.uk, mdean@bbn.com, lynn.stein@olin.edu, hendler@cs.umd.edu, connolly@w3.org, www-archive@w3.org
Enrico, As you might guess, I disagree with what you wrote. Of course, surface syntax should not give false impressions, but I don't think this will be the case here: Some of the typical DAML+OIL idiom corresonds >*exactly*< with the usual frame-based constructions. (for example the locally defined range-striction on a slot, which I used as example in my msg; same for cardinality constraints; same for "defined"-classes; etc). Your point about behaviour requires more thinking, I agree: > Even when dressed as a frame system a DL will always > behave as a DL (with anonymous classes and automatic > re-classification) I know that Stefan Decker has been thinking about a restricted semantics of DAML+OIL which would indeed only deal with named classes (for instance to classify instances). So, in my view, there are two issues, one easy, one hard, both important: - (easy): define frame-based modelling-idiom as syntactic constructions for DAML+OIL - (hard): try to think what DAML+OIL would look like if it only dealt with named classes Frank. ----
Received on Tuesday, 18 December 2001 18:03:11 UTC