- From: Bijan Parsia <bparsia@cs.manchester.ac.uk>
- Date: Wed, 25 Mar 2009 20:46:21 +0000
- To: Oliver Ruebenacker <curoli@gmail.com>
- Cc: Matthias Samwald <samwald@gmx.at>, Mark Wilkinson <markw@illuminae.com>, Phillip Lord <phillip.lord@newcastle.ac.uk>, W3C HCLSIG hcls <public-semweb-lifesci@w3.org>
On 25 Mar 2009, at 20:35, Oliver Ruebenacker wrote: > Hello Bijan, All, Hello Oliver. > These are not even transitive, right? Which? I have 4 or so sketches. Similarity can be transitive, for example. Transitivity isn't the only thing one might want to do. > So what can we do with them > reasoning-wise? Depends on the relation proposed. Thus far we've just sketched things in a very broad way. Clear requirements haven't been developed. I'm not sure why you are so knee jerk against exploring possibilities. > Isn't likeness too much in the eye of the beholder No. Similarity is one sort of possible relation and there are many similarity metrics possible. And many ways of user defining it. Thats sort of the point of having a logic. > to be agreed upon > universally? I regard universal agreement as a non-goal. > Why not move such concepts to more specialized ontologies > instead of OWL? OWL isn't an ontology. We're talking about what might be useful relations to have expressible with standard behavior up to and including inference support. I.e., thinks which are extensions or layers on OWL. Of course, you can axiomitize arbitrary distinct equivalence relationship in OWL since you have Transitivity, Reflexivity, and Symmetry. So that is, of course, another option. It won't work with counting, but perhaps that's ok for some purposes. It should be mentioned as an option of course. Of course, if people don't think investigations along these lines are fruitful, I'm happy to drop it. Cheers, Bijan.
Received on Wednesday, 25 March 2009 21:00:11 UTC