Re: PROV-ISSUE-11 (definition-version): Definition for Concept 'Version' [Provenance Terminology]

> We have so far regarded "Version" as a bad concept. There is no objective
> way to define what differences different versions may have.
> "Version" is a fuzzy notion
> for one of two or more similar objects with a derivation history between them.

IMO, version can either refer to version descriptions or version 
relationships. Descriptions about versioning, such as versioning 
numbers, should be out of the scope of provenance. Statement about the 
versioning relationship between two objects could be part of provenance, 
as a specific type of derivation relationship. But this kind of more 
specific relationship type could be treated as an extension to the core, 
rather than as part of the core provenance concepts.

cheers,

Jun



It is
> a relative concept, and as such does not qualify as class in an ontology, but as relationship:
> "is version of".
> The fact that there is a derivation processes linking two data sets should
> be enough from a Provenance point of view. A "has derivative" link is a deduction
> from the process. One might specify which sorts of processes would be regarded to
> be allowed for derivations that lead to "versions". Only then we may have a reasonable definition
> of "is version of". I regard it however as much too specific for a core Provenance model.
>
> Martin
>
>

Received on Tuesday, 24 May 2011 15:27:13 UTC