Update: There appear to be quite a few items with duplicate Scotland IDs (not all of them may be erroneous!):
http://wdq.wmflabs.org/stats?action=doublestring&prop=709

On Tue, Jun 2, 2015 at 10:23 AM Magnus Manske <magnusmanske@googlemail.com> wrote:

I created (some/most of) these items as part of the Wiki Loves Monuments UK 2014 drive, to run the campaign from Wikidata rather than from a bespoke database. This allows the community (TM) to maintain the data, rather than one poor sod (e.g., myself) having to frantically update all of it every year ;-)

"Consumer" tool is here:
https://tools.wmflabs.org/wlmuk/index_wd.html

These are based on "official" data from National Heritage, provided to me via Wikimedia UK. Grade A (or Grade I/II* in England) structures should be noteworthy by default.

It appears (as per your examples) that some of these were created as duplicates/with wrong IDs. As I said, this is based on "official" data, so it's the best I could do at the time. With mass creation, there are bound to be a few strays. If you can find some large-scale, systemic issue I'll try to fix it, but the one-offs will always fall back to manual fixing. At least, with Wikidata, we can fix them together.

On Tue, Jun 2, 2015 at 10:01 AM Daniel Kinzler <daniel.kinzler@wikimedia.de> wrote:

Am 01.06.2015 um 22:26 schrieb Markus Krötzsch:
> Finally, the technical question is: Why is this even possible? I thought that,
> in each language, label+description are a key (globally unique), yet here we
> have many pairs of items with exactly the same label and description. Or is the
> problem that no description was entered and so the system does not apply the
> key?

The uniqueness constraint does indeed not apply if there is no description.

--
Daniel Kinzler
Senior Software Developer

Wikimedia Deutschland
Gesellschaft zur Förderung Freien Wissens e.V.

_______________________________________________
Wikidata mailing list
Wikidata@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata