- From: Graham Klyne <Graham.Klyne@zoo.ox.ac.uk>
- Date: Wed, 27 Jun 2012 23:23:01 +0100
- To: James Cheney <jcheney@inf.ed.ac.uk>
- CC: Luc Moreau <L.Moreau@ecs.soton.ac.uk>, Provenance Working Group WG <public-prov-wg@w3.org>
On 27/06/2012 18:56, James Cheney wrote: > Yes, that worries me too. I have read over a bit of Guha's thesis. Things get complicated quickly, though (unlike Guha) at least we do not contemplate having the syntax of formulas to depend on the context (if we think bundles are contexts). So, more standard semantic structures from modal logic "ought to work" (famous last words). Yes, I have always tried to avoid that level of complexity. I've thought about fixing interpretations across contexts, but when classes and relations are part of the interpretation that's hard to do completely (cf. IEXT and ICEXT in RDF semantics). > To me, adding something complicated at the last minute is not a good idea (invoking Hoare again). It can always be added later when it is well understood, but if we add something broken now we're stuck with it forever. Maybe a good compromise is to add the syntax (and saying how/why people might use it) without pretending to have a semantics or constraints. Absolutely! This is very much what is driving my current position. And is behind my proposed compromise. #g -- > --James > > On Jun 27, 2012, at 12:24 PM, Graham Klyne wrote: > >> James, >> >> I think I said something in the IRC chat for the face-to-face meeting, but I'll try and reconstruct it here for the record. I think your attempt at formalizing contexts is very useful, if only because it helps to show how tricky a problem this is. >> >> One thing that comes out clearly to me from your semantics is that capturing the semantics of contextualization (at least in your approach, and I suspect in any possible way that addresses the motivating use-case) requires semantics that goes beyond the expressive capability of current RDF semantics. The current RDF semantics has no capacity for dealing with multiple worlds - it depends on a single interpretation mapping that applies to all terms used in an RDF expression (graph), and has no way to link the semantics of disparate RDF expressions. >> >> As I understand it, the RDF dataset structure provides a syntactic framework to support richer semantics, but as far as I can tell there are not yet any actual semantics defined for it. >> >> #g >> -- >> >> On 27/06/2012 16:32, James Cheney wrote: >>> Hi, >>> >>> Neither of those readings below make sense to me... >>> >>> I thought the definition of contextualization before was: >>> >>> An entity that is a contextualization ◊ of another entity presents all aspects of the latter as per the latter's description in another bundle (referred to as remote bundle), and therefore constitutes a particular case of specialization of the latter entity. >>> >>> Your new text in describing the bundle argument to specializationOf just says this: >>> >>> bundle: an optional identifier (b) of a bundle that contains a description of supra and further constitutes one additional aspect presented by infra. >>> >>> and this is saying something completely different to: now the bundle is an aspect, rather than the context that includes aspects that we claim e2 also has. It's not that clear to me what this means (and there is no longer clarifying text about supra being described in the remote bundle b). >>> >>> I understand we want to avoid the word "context", but this seems to be both renaming and changing the meaning (to something new), which goes beyond what I thought we agreed. >>> >>> (I thought we had agreed to rename contextualization to something else, but expected that this would be a separate relation, not overloading specialization. Should have asked.) >>> >>> FWIW, see also http://www.w3.org/2011/prov/wiki/FormalSemanticsWD5#Bundles which attempts to formalize what I thought contextualization was about. >>> >>> --James >>> >>> On Jun 27, 2012, at 9:52 AM, Luc Moreau wrote: >>> >>>> Hi James, >>>> >>>> Assuming we have a "top level" bundle (I am not sure what it would be exactly), >>>> I don't think that the two expressions you are suggesting are equivalent. >>>> >>>> specializationOf(e1,e2,toplevelBundle) >>>> >>>> indicates that e1 shares all aspects of e2 and presents a further aspect: the bundle toplevelBundle. >>>> >>>> >>>> specializationOf(e1,e2) >>>> >>>> indicates that e1 shares all aspects of e2 and presents and further specific aspects (but without >>>> indicating which ones): a bundle MAY or MAY NOT have been fixed. >>>> >>>> Luc >>>> >>>> On 06/27/2012 02:57 PM, James Cheney wrote: >>>>> Hi, >>>>> >>>>> I am happy with renaming contextualization to something less controversial, but renaming it to specialization seems (to me) confusing, unless it's clear that the semantics of the two variants are compatible. >>>>> >>>>> Do we have a name for the "top level" bundle in PROV (i.e., the anonymous bundle that contains the toplevel expressions), and supposing we do, is >>>>> >>>>> specializationOf(e1,e2) >>>>> >>>>> equivalent to >>>>> >>>>> specializationOf(e1,e2,toplevelBundle) >>>>> >>>>> ? >>>>> >>>>> --James >>>>> >>>>> On Jun 27, 2012, at 4:49 AM, Luc Moreau wrote: >>>>> >>>>> >>>>>> All, >>>>>> >>>>>> At the face to face meeting, we have agreed to rename contextualization and mark this feature >>>>>> at risk. Tim, Stephan, Paul and I have worked a solution that we now share with the working group. >>>>>> >>>>>> Given that contextualization was already defined as a kind of specialization, we now allow an optional >>>>>> bundle argument in the specialization relation. (Hence, no need to create a new concept!) >>>>>> >>>>>> See section 5.5.1 in the current Editor's draft >>>>>> http://dvcs.w3.org/hg/prov/raw-file/default/model/prov-dm.html#term-specialization >>>>>> >>>>>> Feedback welcome. >>>>>> >>>>>> Regards, >>>>>> Luc >>>>>> >>>>>> PS. Tracker, this is ISSUE-385 >>>>>> >>>>>> -- >>>>>> Professor Luc Moreau >>>>>> Electronics and Computer Science tel: +44 23 8059 4487 >>>>>> University of Southampton fax: +44 23 8059 2865 >>>>>> Southampton SO17 1BJ email: l.moreau@ecs.soton.ac.uk >>>>>> United Kingdom http://www.ecs.soton.ac.uk/~lavm >>>>>> >>>>>> >>>>>> >>>>>> >>>>> >>>>> >>>> >>>> -- >>>> Professor Luc Moreau >>>> Electronics and Computer Science tel: +44 23 8059 4487 >>>> University of Southampton fax: +44 23 8059 2865 >>>> Southampton SO17 1BJ email: l.moreau@ecs.soton.ac.uk >>>> United Kingdom http://www.ecs.soton.ac.uk/~lavm >>>> >>>> >>>> >>> >>> >>> >>> >>> The University of Edinburgh is a charitable body, registered in >>> Scotland, with registration number SC005336. >> > >
Received on Thursday, 28 June 2012 08:46:14 UTC