Re: Using versionable artefacts in SDMX-ML

Hi Sarven,

Talking for myself and not formally representing the working group then 
I think versioning information for codes and codelists is quite separate 
from Data Cube (nee SDMX RDF).

Just as Data Cube chooses to use SKOS then any versioning information 
should built separately and be applicable to SKOS and other code 
representations, not tied to use within Data Cubes.

Within in the broader RDF world then as well as the OWL annotation 
properties there is dct:hasVersion/isVersionOf, there is the W3C status 
vocabulary [1] and for the UK registry work [2] (which is how at least 
some groups may manage status of codes in codes lists) there is both 
version information (including use of owl:versionInfo) but separately 
status information (see section [3] for more details on that). The 
registry information model does include validity date information and 
the API provides a means to access both the status (of the whole code 
list or individual codes) at a given date or its entire history trace.



On 20/02/13 21:30, Sarven Capadisli wrote:
> Hi,
> If there is a backlog on the following, pointers are appreciated. While
> working on SDMX-ML transformations, versioning artefacts came up. This
> is about using RDF properties which corresponds to them.
> SDMX-ML publishers use @version, and sometimes accompanied with
> @validFrom and @validTo on CodeLists and Codes. I think their
> explanations are self-evident without diving too deep into that for the
> moment. Those attributes and their values are referenced by the
> components in Data Structure Definitions. As classifications evolve over
> time, version and validity information reflects those changes.
> As far as I know, there are no corresponding RDF properties for version,
> validFrom and validTo in SDMX-RDF.
> The best "version" like property that I can think of (and do use) is
> owl:versionInfo. That's fine for now I guess.
> validFrom and validTo is a bit more tricky. It would be nice to have
> them satisfy the following:
> * have range URI (e.g., to use UK refPeriods URI)
> * be datayped with xsd's time periods
> I think the separation of validFrom and validTo is necessary as either
> or both may be available.
> See also [1].
> Would it make sense to have sdmx-concepts for these? Would other
> properties be welcome? :) e.g., isFinal
> [1]
> -Sarven

Received on Thursday, 21 February 2013 08:44:14 UTC