Provenance and monotonicity (was: PROV-ISSUE-1 (define-resource))


I think this is a valid point of view.  Whether it's more "correct" than what I 
was suggesting can only be answered by considering what you want to use the 
provenance for, I think.  So it's back to the use-cases :)


Paolo Missier wrote:
>    James, Graham (will address Luc's comments on the wiki again separately)
> I would argue that the provenance of the car includes all the engine 
> replacements that took place, so if the engine is now B, I would like to 
> ask the question "why is B here?" and receive an answer like "B has 
> replaced A [at time t] [because A failed...]". I believe James hinted at 
> this. And if you are interested, you go back and unfold the history of 
> A. So yes, the provenance of A is still part of the car's provenance, in 
> the car's current state -- the logbook of car repairs that you get from 
> your garage is a simple example.
> All I meant to say is that history is cumulative and immutable. That is 
> not to say it's linear. Someone else (sorry, mail chaos at this point) 
> commented that it is a DAG, and I would agree without having thought too 
> hard (which I never do :-)).
>   The issue of scoping / avoiding the big bang problem is addressed 
> separately: you may decide to prune the early episodes in history for 
> convenience, engineering issues, etc., and for most resources (whatever 
> your definition), there is some kind of origin. It's often relative to 
> the observer (as is all provenance): consumers generally don't need to 
> investigate where the engine's materials come from, whereas a forensic 
> epert investigating an engine failure may.
> In my view, Theseus's ship is the result of all the actions that were 
> ever taken on it, including the destructive ones. Too radical?
> --Paolo
> On 6/2/11 12:41 PM, James Cheney wrote:
>> Yes, these issues seem intuitive only as long as you don't stop to 
>> think about them too hard :)
>> I would say that the provenance has to be scoped by (say) a start and 
>> end time, or some other criterion, to prevent the "big bang" problem 
>> (see e.g. [Miles IPAW 2006]).  
>> If we want the provenance of the car from "now" until it was made, 
>> then the provenance of A needs to be included (e.g., maybe A caused 
>> damage to the car when it failed, so we need to know that to 
>> understand how the car's current state was obtained from its initial 
>> state).
>> If we want the provenance of the car from "now" until I bought it, 
>> which happened after the engine was replaced, then maybe I don't need 
>> to know about A.  (If I want to buy the car, I'd probably value the 
>> knowledge of the earlier history so that I can understand its current 
>> state, but the seller isn't always obligated to provide this.)
>> This reminds me of another good story:
>> /The ship wherein Theseus <> and 
>> the youth of Athens <> returned 
>> [from Crete <>] had thirty oars, and 
>> was preserved by the Athenians down even to the time of Demetrius 
>> Phalereus <>, for they 
>> took away the old planks as they decayed, putting in new and stronger 
>> timber in their place, insomuch that this ship became a standing 
>> example among the philosophers 
>> <>, for the logical question 
>> of things that grow; one side holding that the ship remained the same, 
>> and the other contending that it was not the same./
>> —Plutarch, /Theseus []/
>> /
>> /
>> What is the provenance of the ship?  Was the ship really "preserved"?
>> --James
>> On Jun 2, 2011, at 12:05 PM, Graham Klyne wrote:
>>> I think Paolo has usefully threaded a path through our discussions. 
>>>  Thanks!  At first reading, I would consent (in the sense of 
>>> "consensus") to definitions framed on the basis of what he has 
>>> written here.
>>> ...
>>> The issue of monotonicity (of provenance of a stateful resource) is 
>>> interesting. IntuitivelyProveancne and monotonicity (was: , it seems appropriate, but I'd need to let 
>>> it stew awhile before accepting it unconditionally.  My immediate 
>>> concern is how do we account for correction of previous errors in 
>>> provenance claims?  But this question goes to the heart of what is, 
>>> IMO, one of the key purposes of provenance on the Web (i.e. to help 
>>> deal with conflicting information in the Web, and the Semantic Web in 
>>> particular), so maybe that point gets addressed separately in any case.
>>> Aha!  I just thought of another example:  suppose we're talking about 
>>> provenance of a car (e.g. for QA purposes).  Initially, suppose it 
>>> has engine A, made by a particular factory.  The provenance of the 
>>> car include the provenance of engine A.  Sometime in its life, the 
>>> engine fails and is replaced by engine B, and provenance of engine C 
>>> becomes part of the car's provenance.  At this point, does it make 
>>> sense to claim that the provenance of A is still part of the car's 
>>> provenance?  A similar example could be constructed for, say, a photo 
>>> album where images are added and removed.
>>> #g
>>> --

Received on Thursday, 2 June 2011 21:16:04 UTC