Martin Koppenhoefer <dieterdreist <at> gmail.com> writes: > > > sent from a phone > > > Il giorno 09 giu 2016, alle ore 02:23, Minh Nguyen <minh <at> nguyen.cincinnati.oh.us> ha scritto: > > > > If I understand correctly, you’re referring to a situation where, for > > instance, Wikipedia editors have opted to discuss both an “administrative > > territorial entity” and its government in the same article, whereas Wikidata > > editors have decided to separate the concepts into two different items > > at least the former is very common for cities I believe (didn't conduct a scientific study about it though), > I'm not so sure how common the latter is, my impression when I last looked was that many of the socio > geographic entities are still missing in wikidata (so rather than using 2 distinct objects there is one > which corresponds to a part of the article), their editors seem to have a preference for political > administrative entities.
I’m not sure whether we’re talking about the same concepts, so here are some data points about the kinds of Wikidata items that iD might end up associating with place POIs or boundary relations (among the many kinds of features that can be tagged with `wikipedia` and `wikidata`): * Currently, Wikidata has 1,775,365 items tagged as “human settlements”. [1] You’d expect to see these items tagged on place POIs, including cities and villages but also places that don’t have boundaries (like unincorporated places). * “Administrative territorial entity” is the superset of “human settlements”. This superset has 2,225,880 items. [2] You’d see these items on place POIs and boundary=administrative boundary relations. * “Human-geographic territorial entity” is the superset of “administrative territorial entity” that also includes cultural and purely political boundaries. That superset is less than 1% larger (2,245,631). [3] Something that’s a human-geographic territorial entity but not an administrative territorial entity probably shouldn’t be mapped in OSM in the first place. Rather than searching Wikidata, iD essentially only follows the explicit link from the user-specified Wikipedia article to its Wikidata item. The presence of this link indicates that Wikipedians currently consider the item to be synonymous. To get a better sense of how widespread these iD-generated tags would be, the queries above could be filtered to those items that have sitelinks. Unfortunately, that particular query times out on the Wikidata Query Service. (It’s a great service, but like overpass turbo, it has its limits.) > > Wikipedia tends to be proactive about creating separate articles when > > there’s a notable distinction between the various meanings of a name, but > > Wikidata follows suit almost as a rule. So there is a 1:1 correspondence > > between the various meanings of China on the English Wikipedia and the > > various Wikidata items for those meanings. `wikipedia=en:China` maps to > > <https://www.wikidata.org/wiki/Q148>, which is for the People's Republic. If > > the mapper had a different definition of China in mind, both the `wikipedia` > > and `wikidata` tags would reflect that. > > clearly the china article in wp/en has a much broader scope than the linked Wikidata item people's republic > of china. The latter starts looking at things from 1949, Wikipedia "China" some thousand years earlier. The Wikipedia article provides historical context, including historical borders, as any encyclopedia article on this topic should. By contrast, the Wikidata item is more focused on the present, much the way an OSM boundary relation should reflect the present. (The historical boundaries go in OpenHistoricalMap, after all.) Both Wikipedia and Wikidata have entries for other definitions of China. [4] To me, this suggests that the Wikidata item more naturally fits an OSM feature than the Wikipedia article does. [1] http://tinyurl.com/h9p4qb7 [2] http://tinyurl.com/zvrft74 [3] http://tinyurl.com/zuzcvly [4] https://en.wikipedia.org/wiki/China_(disambiguation) and the Wikidata items linked to the articles linked from that page -- Minh Nguyen <m...@nguyen.cincinnati.oh.us> _______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk