- From: Graham Klyne <Graham.Klyne@zoo.ox.ac.uk>
- Date: Thu, 28 Jun 2012 12:25:29 +0100
- To: Luc Moreau <L.Moreau@ecs.soton.ac.uk>
- CC: James Cheney <jcheney@inf.ed.ac.uk>, Timothy Lebo <lebot@rpi.edu>, Satya Sahoo <satya.sahoo@case.edu>, Provenance Working Group WG <public-prov-wg@w3.org>
On 28/06/2012 09:12, Luc Moreau wrote: > Hi James, > > As far as your suggestion is concerned, this is mostly how Tim mapped > Contextualization to prov-o. > Some trade-offs are necessary. We can discuss them. > > However, I don't think it solves Graham's problem! Actually, I think the proposal is not so far from an earlier proposal I made, which is to keep specializationOf and inBundle relations separate. You objected to this because it led to conclusions like an entity being both slower *and* having good performance. If I understand it correctly, James's proposal here would lead to similar conclusions (which I think he acknowledges by saying "I don't see how this addresses the original motivation for ctxOf"). My problem is that I don't think you can avoid that conclusion without a proper semantics of context, which RDF does not currently have. #g -- > On 06/28/2012 04:35 AM, James Cheney wrote: >> If all that we need is a way to be able to say "e2 is a specialization of e1 >> which is described in bundle b", then would the following suffice: >> >> 1. Leave out ctxOf (or the 3-ary form of specializationOf) >> 2. Add a special attribute "prov:inBundle" that any entity (or indeed anything >> else) can have, linking each entity id to a bundle it appears in (there may be >> more than one). >> 3. Then instead of ctxOf(e2,e1,b) we just say specializationOf(e2,e1). The >> fact that e1 happens to be in bundle b gets transferred by e2, along with all >> other attributes. (Which seems weird to me, since e2 isn't explicitly >> mentioned in b, but if being in a bundle is just an ordinary attribute, then >> it should be transferred by specialization just like every other attribute). >> >> I don't see how this addresses the original motivation for ctxOf, but don't >> see that it does any harm - the complications arise if we start trying to >> assign different meaning to "entity e" and "entity e in bundle b". >> >> Just putting this out there - I am not pretending to understand that I >> understand what ctxOf means at this point, and so the probability that I'm >> barking up the wrong tree is high. But maybe finding out why this is wrong >> will be educational. >> >> --James >> >
Received on Thursday, 28 June 2012 11:32:18 UTC