[wdmmg-discuss] Where are the Government Departments?

Sun Aug 15 01:40:59 UTC 2010

Here are two poorly researched ideas about Government departments:

1. I spoke to Chris Taggart about coding up departments when I met him at 
OK Con. He seems to have a pretty clear picture of the prior work in 
this area, and all its deficiencies. More importantly, he said he has 
invented a coding system for his own use, and he encouraged us to use it. 
I never followed this up in the end, but it might be an avenue worth 
exploring.

2. Last time I checked, we were quite close to having a decent list of all 
the Programme Objects. These are organised into a hierarchy. The next 
coursest level is Programme Object Groups, which appear in the CRA. After 
that there is at least one more level, and then the top level is the 
department that allocated the code, as I understand it. If so, then you 
might be able to get a usable de facto coding system out of it. Compared 
to other coding systems, it would have the advantage that any department 
that allocates PO codes will have an incentive to maintain it. I bet it 
therefore somehow copes with the historical awkwardness you describe 
below.

I could be completely wrong. :-)

 	Alistair

On Sat, 14 Aug 2010, William Waites wrote:

> I'm having a look at making an SDMX/RDF representation
> of the Country and Regional Analysis, one of the main datasets
> behind the Where Does My Money Go project.
>
> One of the dimensions is government department, and it would
> be nice to be able to link to the canonical concept scheme. I
> believe there was some work going on on this back in the spring
> and I wonder what has become of it.
>
> Looking at the most recent dump in [0] there seem to be a
> number of problems:
>
>   * There is no schema. There is a suggestion of a concept
>      scheme but [1] gives a 404
>   * none of the URIs dereference, it appears not to have been
>      loaded into the Talis endpoints
>   * The data doesn't include historical (even recent) departments.
>      For example, the Department for Innovation, Universities and
>      Skills was merged into the Department for Business, Innovation
>      and Skills relatively recently and appears in Treasury
>      documents covering the period of interest.
>   * Related to the last, there is no mention of a creation and
>      winding up date for the departments, nor any notion of
>      predecessors and successors.
>   * Currently extant departments, for example HM Revenue and
>      Customs do not exist (thought that one is mentioned in the
>      comment for the attorney general).
>
> There was a bunch of very good work done back in January
> and February by Stuart Williams that addressed most of these
> problems. I can see the data on the Basecamp. The dumps on
> source.data.gov.uk seem to have removed a lot of information.
> There was a schema, date and lineage linkage, all of that.
>
> Stuart's that data wasn't necessarily perfect, it collated information
> from a large number of sources. But was is far better than what we
> have which is more or less nothing. Errors can be corrected but
> they'll be much more easily discovered if the information is open
> to inspection and use.
>
> It's an iterative process, even for the government. In this particular
> case, we discovered that there is actually no centrally held canonical
> list of public bodies, this is one of the reasons that Stuart's work
> was so difficult and valuable. Simply admitting this and putting
> a caveat that for this reason the data may contain errors and is
> subject to revision would be a lot better than waiting until it is
> perfect to publish the data.
>
> To practical questions: can I use this unpublished data? Can I
> publish it and make it available in my SPARQL endpoints
> which are obviously not run by *.gov.uk?
>
> [0] http://source.data.gov.uk/data/reference/ministerial-departments/
> [1] http://reference.data.gov.uk/def/public-body/Department
>
> -- 
> William Waites <wwaites at gmail.com>
> Mob: +44 789 798 9965
> Fax: +44 131 464 4948
>
>
> _______________________________________________
> wdmmg-discuss mailing list
> wdmmg-discuss at lists.okfn.org
> http://lists.okfn.org/mailman/listinfo/wdmmg-discuss
>