Proposal talk:Wikipedia tag types

one wikipedia tag is enough

I would prefer to distinguish the three different ways of linking to Wikipedia - to an article, to a part of an article, or to a list containing something - in the *value* space and not in the *key* space. After all, it would not make sense to have an OSM object having both a Wikipedia link and a wikipedia_list tag (as the proposal itself says - if a proper Wikipedia article comes around, the list reference could and should be scrapped). Therefore, I would like to have just one wikipedia key and then the value could be something like "list:name_of_wikipedia_list" or "section:name_of_main_article" or something like that. --Frederik Ramm (talk) 13:01, 4 March 2017 (UTC)

Frederik Ramm, the format of the value has been very well established at this point -- language_code:normalized article title", and has been around for a while. If we all of a sudden introduce a new format of the value, most of the data consumers will incorrectly assume that "list:" is a name of the language, and break/ignore/fail. If possible, it would be good to avoid "magic hacks" like that, as it makes all consumer code much more fragile. "key" is a contract, it defines the exact meaning and the format of the value, and should not be changed if we can avoid it. If we conflate multiple meanings into the same key, we introduce a potential for problems.

Also note that we already use a different key for the list/section usecases (see wikipedia secondary links), you see that community uses a wikipedia:language=title instead of wikipedia=language:title in cases of section and lists. I think the conflation here is not clear to consumers and documenters alike, and does not convey the meaning of the key fully. --Yurik (talk) 18:55, 4 March 2017 (UTC)

Thank you Yurik you said that better than I could have. Keithonearth (talk) 08:11, 16 March 2017 (UTC)

I'd second Frederik's argument not to change the key but the value and I also see use of current wikipedia value. Why not just promote using of #section wikipedia values? for cases when the object is just mentioned (e.g. "Rapperswil#Tourist_attractions" for Rapperswil Castle, or "de:OpenStreetMap#Lizenzen")? --Geonick (talk) 14:57, 19 March 2017 (UTC)

Most wikipedia articles are not a 1:1 fit with the object

Most wikipedia articles about places include both, the concept of administrative entity and socio-geographic place (and focus a lot on the administrative entity part, usually it's in the first sentence). A lot of wikipedia articles speak about different items or contain references and descriptions to things that might be mapped individually. E.g. an article about a church might reference different building parts, different functional parts like chapels, artwork like paintings and sculptures, historic items like tombs, tombstones and sarcophags. An article about a museum will usually also cover the buildings it is in, etc. Where do you draw the line? --Dieterdreist (talk) 13:42, 9 March 2017 (UTC)

And btw.: Wikidata is not (yet?) a solution, it is in the overwhelming majority of cases a 1:1 reproduction of the articles (every article is a wikidata item, that's how wikidata started)., e.g. if a wikidata object represents a wikipedia article about a settlement, which typically also relates to an administrative entity, the wikidata item will also say the thing is an administrative entity and a settlement. In wikidata it is defacto impossible to make edits like adding properties, because you would have to check with all languages, and you can easily introduce errors without noticing. See [1] the object for "city". In German this is "Stadt", but the description and definition for Stadt (as for "città in Italian) has nothing to do with an administrative entity. In English, the definition for "city" says: "type of municipality" hence it seems logical that there is also the statement subclass of "administrative territorial entity" and "political territorial entity", which is orthogonal to what the German and Italian wikipedia describe. And even if now everything seems fine, tomorrow someone can come and add another statement to "city" and will change the definition for all the entities that are tagged as "subclass of city", in all the languages, and she will very likely base her edit only on the text in one of these languages and on the situation she is aware of. --Dieterdreist (talk) 14:01, 9 March 2017 (UTC)

Dieterdreist, for the church example, while you can draw multiple ways to represent different part of the church, I think there should be only one relation or an outer way that represents the whole 1:1 concept of the church as described in WP. We could come up with some tag to represent "covered-in-article" (similar to section, but does not link to specific article's portion), but this is outside of the scope for this proposal.

I agree that there are some cases when multiple ways/relations should link to the same Wikipedia article. We could introduce multiple Wikidata entries whenever historic/social and administrative definition of a place are different, but I doubt too many people would write multiple articles. Lastly, for the reasons you describe, we will never achieve a perfect 1:1, but with this proposal we could clarify a vast majority of cases that already exist. So yes, if you try to model every part of the museum, WP/WD tagging gets confusing, but 99% of the times you will have a simple square to represent it, and a two line mentioning of the museum in a town article. Lets solve the majority of cases, and come up with better tags when it really becomes a big problem. --Yurik (talk) 01:13, 15 March 2017 (UTC)

Sections can often be found in several articles

Do you suggest multiple values to tag different sections in different articles referring to the same object? --Dieterdreist (talk) 13:44, 9 March 2017 (UTC)

Semicolon is a valid character in Wikipedia articles, so allowing that would create a potential for problems. I think _1, _2, ... tags should be used instead. I feel the section tag would really highlight the fact that it would be nice to have a full wikipedia article about the subject, but one is not yet there, so in the mean time, link to a section. So the goal is not to link to every section in WP that matches, but just have a starting point. But of course, this is up to contributors. --Yurik (talk) 00:58, 15 March 2017 (UTC)

Do we really want to tag things in Wikipedia lists?

Can't you get the list from wikipedia? Either there is an article for the object (can be linked and will have a reference to the list), or there isn't and a link is of few use anyway. --Dieterdreist (talk) 13:44, 9 March 2017 (UTC)

Dieterdreist, I might agree that there is usually not that much use for lists, but this proposal simply reflects the fact that there are already thousands of objects that point to list articles. Some of the more obvious ones I posted here, but that list is much longer - at this point I list only those that are "instance-of" Wikimedia list article. So we could delete them all, which may potentially anger the contributors, or use a different tag. On top of this, having a list ref may still be benefit - as more articles are written, they may be linked from the list article, making them easier to find. --Yurik (talk) 00:52, 15 March 2017 (UTC)

If there's no 1:1 relation with Wikipedia, use Wikidata

Wikidata distinguishes concepts more accurately, and it should probably be seen as more authoritative in that sense. Linking a Wikidata item can probably solve the problem that #section is trying to fix. Regarding lists, I really can't see a good reason to have those... Pizzaiolo (talk) 20:40, 16 March 2017 (UTC)

Pizzaiolo, I agree that Wikidata is better in that sense, but should we keep wikipedia tag that points to a good article but that only has a small portion about the object? Some "botopedias" automate article creation. They inject a lot of "stub" articles, for all sorts of small places. So there could be a Wikipedia article in some less read language about an object, but not in the language of the land or a more common language like English/Spanish/French/... Should the wikipedia tag be changed to match wikidata tag, and point to a machine-generated article? Should wikipedia tag be deleted? I propose we change it to wikipedia_list. The same logic applies to the section. --Yurik (talk) 05:00, 17 March 2017 (UTC)