Topic on User talk:Jdforrester

Hi James,

I've been wondering what to do with Member of Parliament for Pudsey (Q16707840). As far as I can see you created this as an example of what could be done with the officeholder property. However, we've since gone down a different route for MPs (with constituency as a qualifier rather than subdividing the item). As far as I can see no others like this were ever created, and the only two items using it also have the same information expressed in a more standard way. Would you object if I listed it for deletion once I've cleaned those up? I normally would leave surplus items like this be, but I think it might cause confusion if we kept it around despite not using this approach to modelling the data.

Hmm.

The problem with that modelling is that it makes it impossible to represent the successor/predecessor-for-each-part-of-the-constituency issue when constituencies change. For example, representing that George Cunningham for Islington South West mostly-ish "retained" his seat when it became Islington South and Finsbury, but that at the same time he replaced Ronald Brown (as MP for Shoreditch and Finsbury). It also makes events related to the holding (e.g. appointment to the Manor of Northstead, or successful prosecution rending one ineligible); I'd imagined that we might want to represent the "caused the termination of" link from office to office or case to office.

My aspiration with having items for the office was to represent this richness, but as you say I never undertook the full-scale implementation. At the time it wasn't possible to attach geoshapes to claims, so some of the value was not open to us. We could delete it, or we could choose to have these as parallel modelling, though that seems messy.

What do you think?

I've been wondering about that. My long-term plan is to express a before-after relationship between the constituencies themselves - so Shoreditch and Finsbury replaced by Islington South and Finsbury [and somewhere else]. So a query could say "find the seat, find the seat it replaced, find the holder(s) of that seat in the previous Parliament".

This of course needs suitable data at constituency level. At the moment we have one item for every named seat, but they're a bit confusing because many seats had two or three instances under the same title (eg East Surrey, 1832-1885 and 1918-date) - ideally we'd have two separate items for those with separate start/end dates, separate replaces/replaced, etc. I'm getting close to importing all the date/constituency pairs from Historic Hansard, and once those are in, manually splitting up the constituencies will become a necessary next step.At that point adding in some kind of replaces/replaced link will be a natural step (as will shapefiles, if we can get them)

In terms of termination causes, this is something I do also want to do more with - at the moment we mostly use "resigned", "died", "was kicked out", but it could easily be extended to subdivide these further.

Would this approach cover your intended uses?

Plausibly. There's the question of whether it'd be reasonably possible to execute any such query.

For example:

Given an institution A, with a series of head quarters locations Bn, can you get a list of MPs Mn who've represented the constituency in which the institution has resided?

(Or more likely, people with residences, but that's also less likely to be recorded.)

This would be

• A <predicate> Bn; Bn <geosearch> Cn (historic constituencies); Cn <reverse predicate?> Mn with your model, or

• A <predicate> Bn; Bn <geosearch> Cn (historic constituencies); Cn <predicate> Dn (offices of constituency holders); Dn <predicate> Mn with mine.

There's also the parallel MPs for pre-Reform constituencies, of course. :-)

Yeah, this could reasonably work, I think. Of course there's a lot of data filling in to do; George Galloway is a beautifully rendered example to be starting with. Nice work. :-)