- From: Graham Klyne <GK@ninebynine.org>
- Date: Mon, 26 Sep 2011 08:42:56 +0100
- To: "Cresswell, Stephen" <stephen.cresswell@tso.co.uk>
- CC: "Myers, Jim" <MYERSJ4@rpi.edu>, Paolo Missier <Paolo.Missier@ncl.ac.uk>, public-prov-wg@w3.org
On 25/09/2011 22:03, Cresswell, Stephen wrote: > (2) > One use for the original IVPof was (I thought) to relate together long-term entities (e.g. Luc-over-his-lifetime) with shorter-term entities describing states (e.g. Luc-in-Boston). Now it seems that the strongest assertion that I can make about the relationship of these two entities is: > > wasComplementOf( Luc-in-Boston, Luc-over-his-lifetime ) > > ... but this just asserts that Luc's visit to Boston *overlapped* with his lifetime, which is weaker than what I wanted to assert. > > If I also want to describe a visit to MIT that Luc made while in Boston, I could also assert > > wasComplementOf( Luc-at-MIT, Luc-in-Boston ) > > Since the assertions are quite vague, we can't infer that Luc-over-his-lifetime contained Luc-at-MIT, and we can't even infer that they overlapped. > I think it would be useful to be able to make some stronger assertions that allow transitivity to be used here. At some point during its evolution, IVPof was close to being that helpful transitive relation, but now its gone. I think we still need it. +1 ... FWIW, I think my earlier claim: (B wasComplementOf A) <=> exists C.(C IVPof A & C IVPof B) is not correct. Rather: (B wasComplementOf A) => exists C.(C IVPof A & C IVPof B) Because under (my understanding of) the original IVPof, there is no requirement for two IVP values to overlap. #g --
Received on Monday, 26 September 2011 08:46:53 UTC