[ckan-discuss] Package Relationships - remove?

[David Read]

On 2 September 2011 10:02, Paul Hermans <paul at proxml.be> wrote:
> Hi all,
> From a discussion yesterday with a client.
> They do very large surveys and as such they have a dataset covering all
> survey results.
> Later on this large dataset is split up in a multitude of smaller datasets
> using breakdowns on some of the dimensions covered (enterprises, households,
> age, gender, year, …).
> Both the large and more granular datasets are published.
> In one way or another the relationship between the two should be indicated
> to the ckan user.
> Is this a use case for reviving package relationships or should one handle
> this with other features (such as groups ...) ?

Paul,

I'd be tempted to have the shards of data as different resources, all
in the same package. It sounds like the majority of the metadata
(subject, source, area covered, dates covered, date updated) would be
the same, hence having it all in one package (which is a commonality
of metadata).

On the other hand, something like a census would probably be better
split up into different packages, because the subjects covered
(housing, jobs, demographics) can be so varied, that the metadata
could feel quite different for the various shards. In Britain, the
original personal linked data is not released for a few decades, but
the depersonalised shards are published, so this would be v. usefully
be shown as 'derived' from the (closed) original data. So yes, another
good use case.
Cheers,

David

>
> Paul
>
> _______________________________________________
> ckan-discuss mailing list
> ckan-discuss at lists.okfn.org
> http://lists.okfn.org/mailman/listinfo/ckan-discuss
>
>