- From: Timothy Lebo <lebot@rpi.edu>
- Date: Fri, 2 Dec 2011 17:33:36 -0500
- To: Luc Moreau <L.Moreau@ecs.soton.ac.uk>
- Cc: public-prov-wg@w3.org
Luc, On Nov 21, 2011, at 11:12 AM, Luc Moreau wrote: > Hi Tim, > Responses interleaved. > > On 11/21/2011 03:26 PM, Provenance Working Group Issue Tracker wrote: >> PROV-ISSUE-157 (TLebo): wasInformedBy's non-transitivity >> >> http://www.w3.org/2011/prov/track/issues/157 >> >> Raised by: Timothy Lebo >> On product: >> >> The argument that wasInformedBy is not transitive is not clear and convincing. >> >> 1) Does the diagram correctly illustrate the assertions wasInformedBy(a2,a1) and wasInformedBy(a3,a2), and wasInformedBy(a3,a1)? There seems to be a lot of additional unstated assumptions that are embodied by the diagram beyond these three DM assertions. >> > > Hmm .. which? Whatever assumptions were used to map the three records wasInformedBy(a2,a1) and wasInformedBy(a3,a2) and wasInformedBy(a3,a1) to this collection of ascii characters: >> ------ a1 >> | >> e1 >> | >> ------- a2 >> | >> e2 >> | >> ----- a3 Start times, generation times, and usage times are implicit here. a2's e2 usage time is the same as a3's e2 generation time? Time is to the left? A2's start is afer A3's start? The diagram is following unstated conventions. I'm sure my attempt [1] is violating similar assumptions :-/ [1] http://dvcs.w3.org/hg/prov/raw-file/042a0ed56b4d/diagrams/activity-ordering-records-information-flow-ordering-not-transitive.png > >> 2) It appears that the counterexample is using a "degenerate" situation, when the argument for non-transitivity can ALSO be made for activities following a more natural time flow (e.g., even when e2 is used AFTER e1 is generated). In this situation, we do not _know_ that the information in e2 is related to the information in e1 _in any way_. >> > > I don't think we have made a transivity claim in the example you suggest (e2 used after e1). You're making a non-transitiivty claim with the diagram. Even the latest DM draft states the following above the ascii diagram: "The relationship wasInformedBy is not transitive. Indeed, consider the records wasInformedBy(a2,a1) and wasInformedBy(a3,a2), the record wasInformedBy(a3,a1), may not necessarily hold, as illustrated by the following event line. " > > The point, which I would like to adopt across the whole document, is that I would like to explain non-transitivity property for > all relevant relations. That'll be great. > >> 3) There is a "Note: This relation to be simplified using wasStartedBy/wasEndedBy.". Could someone point me to the expected changes here? >> >> >> > Sorry, it was a note to me, it is the following relation wasScheduledAfter that needs simplifying. Great. I very much prefer your simplification and renaming. Regards, Tim > Luc > >> Thanks, >> Tim >> >> >> Reference: >> >> >> http://dvcs.w3.org/hg/prov/raw-file/default/model/ProvenanceModel.html#record-OrderingOfActivities : >> >> """ >> The relationship wasInformedBy is not transitive. Indeed, consider the records wasInformedBy(a2,a1) and wasInformedBy(a3,a2), the record wasInformedBy(a3,a1), may not necessarily hold, as illustrated by the following event line. >> >> ------ a1 >> | >> e1 >> | >> ------- a2 >> | >> e2 >> | >> ----- a3 >> >> The end in activity record identified by a3 precedes the start in activity record identified by a1, while interval for activity record a2 overlaps with each interval for a1 and a3, allowing information to flow (e1 and e2, respectively). >> """ >> >> >> >> > > -- > 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 > > >
Received on Friday, 2 December 2011 22:34:34 UTC