- From: Jim Hendler <hendler@cs.rpi.edu>
- Date: Wed, 20 May 2009 11:52:18 -0400
- To: Ian Horrocks <ian.horrocks@comlab.ox.ac.uk>
- Cc: "OWL 1.1" <public-owl-wg@w3.org>
Ian - I think these are all necessary, but I think there are some others we need for QL and RL - particularly since we claim these to be related to RDBMS technologies, I think we need to see that validated. But please note the following (also addresses what Sandro stated about OWL Full in his email) -- in WOWG we had tests in our test set that tested every single feature (i.e. each possible vocabulary item) added to RDF(S) by OWL - so that the criteria that required every test be passable meant that every feature of Full (and also of the others) was at least separately implementable. The 80% was not meant to show that many features could be supported together - and it also included some where the expectation was that a DL reasoner would not be able to properly handle (and there were a lot of details in the test set about what these meant - see the OWL 1 test case document) I think Michael or someone needs to go through the current OWL 2 design vs. the tests, and make sure we have something in the CR results that gaurantees that every feature is implementable - the 80% should then also be included (or something similar) for the same reason. So I think Ian provides a great starting place - but as the original OWL was only differentiating 3 things, and the new one at least 5 (Full, DL, RL, QL, EL) we have some extra responsibilities. Also because the rationales for some of the profiles are different than the ones in OWL 1 (particularly, where QL and RL refer to RDBMS technology) we must show that the claimed benefits are realizable in implementations. So these are what I would hope we would work towards. -JH p.s. Would also be useful as an internal document to do what we did with the OWL 1 criteria and identify current candidates we think can handle what is being proposed - based on my experience from WOWG, when we presented the Exit Criterion to the Director, he wanted some evidence that our proposed time frames were realistic - by having candidates we were able to seriously shorten the time (if I recall correctly, without that we would have been forced to have at least a 6 month CR) On May 20, 2009, at 11:39 AM, Ian Horrocks wrote: > After discussions with Alan and Sandro I redrafted the "candidate" > CR exit criteria [1] in a way that I hope meets some of the concerns > that have been expressed. Hopefully they are at least a sensible > starting point for our discussion. > > Ian > > [1] http://www.w3.org/2007/OWL/wiki/CR_Exit_Criteria > > "Con un poco de semántica ya se consigue ir muy lejos" Prof James Hendler http://www.cs.rpi.edu/~hendler, @jahendler, twitter Tetherless World Constellation Chair Computer Science Dept Rensselaer Polytechnic Institute, Troy NY 12180
Received on Wednesday, 20 May 2009 15:52:59 UTC