On 12 August 2014 22:49, Andre Engels andreengels@gmail.com wrote:
On Tue, Aug 12, 2014 at 6:26 PM, Sezgin Sucu sezsucu@gmail.com wrote:
https://www.wikidata.org/wiki/Q147630 I realized that, it is labeled as "Zara, Turkey" and not Zara. This seem wrong to me, since "Turkey" part of this label is already in the data as a country relation. It appears to be a one-to-one mapping of the wikipedia article's title. Maybe when first imported from wikipedia, its title is used as is. I want to verify my understanding that labels of entities should be as short as possible and should not attempt to disambiguate as it has to be done in wikipedia article titles.
Again, you seem to be fighting a rule that does not exist. That the title is "Zara, Turkey" is indeed because it is copied from the Wikipedia article by that name, but Wikidata welcomes human as well as automated edits, and I do agree that just "Zara", with Turkey mentioned in the description, not the title, would be better.
Agree. I believe a lot of the importing did try and remove disambiguation - however, they restricted themselves to cases like:
WP: Zara (Turkey) > WD label: Zara
In this case, it's obvious that everything after the comma is a disambiguation note, but there are many articles where the commas form part of the name and so you can't trim automatically - a good example is "Ross, Skye and Lochaber (UK Parliament constituency)", which should definitely be "Ross, Skye and Lochaber" not "Ross" :-)