UK: Wrong GSS Codes for constituencies

[ajparsons]

Problem with Existing Data

Legislature

UK Parliament

Problem

GSS codes used refer to the pre-2010 constituencies and have expired. These are of course valid for 1997 to 2010 in constituencies that haven't changed their name - so cannot be removed. Means EP data can't be used to reconcile datasets based on area.

Proposed Solution

Migrate current GSS to 'GSS_1997' (future-proof against future reviews).
Populate 'GSS_2010' - all in EC results release but also seems to be in wikidata. http://www.electoralcommission.org.uk/__data/assets/excel_doc/0011/189623/2015-UK-general-election-data-results-WEB.xlsx

[tmtmtmtm]

@ajparsons all the GSS codes in EP come from Wikidata, so this will need to be fixed upstream. Can you give an example?

[ajparsons]

It's be all of them that existed prior to 2010. Looks like the problem is on our end because the wikidata item for Cardiff North for instance contains both and knows the expiration date. https://www.wikidata.org/wiki/Q1077332

We need to get both those codes in in some way.

[tmtmtmtm]

There are a few different issues here.

The first is that the Wikidata items have both codes as being of 'normal' rank, when the current one should be marked as 'preferred' (which I've done for the Cardiff North example at https://www.wikidata.org/w/index.php?title=Q1077332&type=revision&diff=609552471&oldid=604918798)

When that gets done, EP will at least end up with the correct current value.

We could probably automate that relatively easily if everypolitician/position_statements#15 were done.

Handling historic information is trickier. I think the correct solution there would be to split the Wikidata item in two, such that the pre-2010 constituency is a separate item from the post-2010 constituency. Then each can have the correct GSS code, boundary information etc attached, and the memberships can be pointed at each. This has already happened in some cases where changes have been more substantial, so this would a matter of applying that more widely.

Doing that in an automated way would require a little bit more thought.