Talk:Q15916867

Classification

@Popcorndude: I restored my version, per Wikidata_talk:Country_subdivision_task_force#Australia.

@TomT0m: But the many items that are subclasses of this are then NOT subclasses of administrative territorial entity (Q56061), resulting in constraint violations in many places.

Also, administrative territorial entity of a single country (Q15916867) represents administrative territorial entity (Q56061)s in a particular country, so every administrative territorial entity of a single country (Q15916867) is necessarily a administrative territorial entity (Q56061) and the description of subclass of (P279) says "all of these items are instances of those items; this item is a class of that item", so shouldn't this then be a subclass of administrative territorial entity (Q56061), perhaps in addition to administrative territorial entity type (Q15617994)? Popcorndude (talk) 21:36, 30 June 2015 (UTC)

@Popcorndude: Please continue this discussion on the administrative subdivision project, and read especially the table with 3 column and the discussion with Filceolaire. As a complementary reading also Help:Classification should be some help. But I can't replicate all of these here, I'm tired of that. TomT0m ^(talk) 08:12, 1 July 2015 (UTC)

@Nikki, Popcorndude, El dizino: up.

Which constraint exacty is the problem ? We should correct it, not build a weird model. Let's discuss how to make things consistent. TomT0m ^(talk) 16:48, 2 July 2015 (UTC)

I re-added subclass of (P279) administrative territorial entity (Q56061) to administrative territorial entity of a single country (Q15916867) because administrative territorial entities of a single country are a subset of all administrative territorial entities and because you created thousands and thousands of constraint violations across a wide range of properties by removing it. I don't understand what you're trying to achieve here. - Nikki (talk) 16:57, 2 July 2015 (UTC)

@Nikki: I created this item, I know what I wanted to do and it's not that. an the explanation is on Wikidata Talk:WikiProject Country subdivision § Australia. In short : Paris is not a type of administrative division. It's just an administrative division. French city on the other hand is a type of administrative divisions. TomT0m ^(talk) 17:06, 2 July 2015 (UTC)

@Nikki: And if you want to understand what I'm tying to achieve (having a classification system which is flexible, stronrg and well founded, please read Help:Classification. After that you're welcome to comment on Adopt Help:Classification as an official help page. TomT0m ^(talk) 19:00, 2 July 2015 (UTC)

@TomT0m: Ohh, I think I see what the problem is: Looking at the history, it was Infovarius who created this item (with Russian and English labels) and you later added a French label. The French label appears to mean something different to what the English means - that's probably the root of the confusion here, because I agree that Paris is definitely not a type of administrative territorial entity of a single country, but it is an instance of an administrative territorial entity of a single country.

Looking at the existing data, almost all of the existing data is using it with the English meaning (WDQ is lagging by over a day, but I assume there weren't that many changes):

• There are 223 subclasses, all of which look like they are valid subclasses of administrative territorial entity (Q56061).
• There are 266 instances, which all (with the exception for administrative territorial entity of Greece 1997–2010 (Q11017753), administrative territorial entity of Republika Srpska (Q19849064), Indian reservation (Q20203616), administrative territorial entity of Australia (Q4494320) and department of France (Q6465)) look like they should be instances of administrative territorial entity (Q56061).

Unless I'm missing something, it seems like the easiest solution would be to create a new item for "type of administrative territorial entity of a single country" (that really means "type of" this time! :)) and fix the French label and the few things which are using this with the "type of" meaning.

- Nikki (talk) 21:20, 2 July 2015 (UTC)

It's also possible that there was a wrong merge in the history, I had no luck (and a Tamawashi) in my enterprise. :) I'll try to fix this. TomT0m ^(talk) 07:44, 3 July 2015 (UTC)

@Nikki: "département français (Q6465)) look like they should be instances of subdivision administrative (Q56061)." Absolutely not. "French department" is a type of administrative division, so it should be a subclass of administrative territorial entity (Q56061). Like paris is a(n instance of) french city, and is at the same time a division of france, of course. This means that cities and department are special kind of. This is the semantics of "subclass of". TomT0m ^(talk) 16:51, 5 July 2015 (UTC)

@TomT0m: I think you misunderstood what I said. By "with the exception of [...] department of France (Q6465)" I mean that (of the 266 "instance of" uses) I see 261 individual entities and 5 types of entity, and department of France (Q6465) is one of the 5 uses which is a type of entity. - Nikki (talk) 22:50, 5 July 2015 (UTC)

@Nikki, Infovarius: Anyway "administrative division of a single country" does not male much sense :) Any administrative division is one of a single country, pretty much. This is replicating the nonsense of the category system of Wikipedia. Actually "type of administrative division of a single country was intended to be the class of all classes such as "french administrative division" "english administratie division", ... TomT0m ^(talk) 16:46, 5 July 2015 (UTC)

@TomT0m:, you are saying strange things. administrative territorial entity of a single country (Q15916867) was intended not as type (class of classes) but as the root (upper class, P279) for all like administrative territorial entity of Greece 1997–2010 (Q11017753), administrative territorial entity of France (Q192498), administrative territorial entity of the United States (Q852446) and so on. Sounds weird in Russian (but may be OK in English and French), Paris (Q90) is modelled now as an instance of administrative territorial entity of France (Q192498). Thus administrative territorial entity of a single country (Q15916867) becomes the root of all administritive divisions in the world, which is convenient. So I was bold and change French label in it. --Infovarius (talk) 16:50, 5 July 2015 (UTC)

In Russian the item means "administrative divisions by (per) country", may be to change French/English labels according to that? --Infovarius (talk) 16:52, 5 July 2015 (UTC)

It's replicating the problems of the Wikipedia categorys system. It's at the "type of administrative division" that this "per country" stuff should be managed. Please see the discussion on Wikidata Talk:WikiProject Country subdivision § Australia. TomT0m ^(talk) 16:57, 5 July 2015 (UTC)

@Infovarius: "administrative divisions" is enough as a root, is'nt it ? TomT0m ^(talk) 16:57, 5 July 2015 (UTC)

Enough. But it is more convenient to separate "by country" to different root from others (city, municipality and so on) mentioned below by Nikki. --Infovarius (talk) 09:07, 6 July 2015 (UTC)

See [1] - While the majority of administrative divisions are going to be an administrative territorial entity of a single country, there are a few which are administrative territorial entity of more than one country (Q15646667), "by country" forms a different subtree than administrative territorial entity of a specific level (Q1799794), and there are also generic terms like city (Q515) which are neither administrative territorial entity of a specific level (Q1799794) nor "administrative territorial entity of a single country". - Nikki (talk) 22:50, 5 July 2015 (UTC)

OK, now I think about it, maybe Type of administrative divisions of a single country should be a in the 4th level ... I'm not sure. TomT0m ^(talk) 17:05, 5 July 2015 (UTC)

I'm not sure what you mean by "should be a in the 4th level". - Nikki (talk) 22:50, 5 July 2015 (UTC)

OK, at that point I'll copy/paste the table of the other discussion

Levels of classification and examples

concrete divisions	concrete divisions classification	type of division classification
Victoria, Queensland, Brisbane (Q34932)   , ...	State, Territory, city, ...	administrative territorial entity type (Q15617994), type of administrative division in australia, ...

@Infovarius: It's easier to separate the roots by putting your more convenient to in the third column. This means that items in the second column will be instance of items in the third column, but nether subclass of them. With this, we have a clear and well founded classification. Do you agree ? TomT0m ^(talk) 09:32, 6 July 2015 (UTC) @nikki: By in the fourth colum, I wonder if we should add a level into the concrete administrative entity > administration entity types > class of administrative entities > ? scheme.

For example I think that we should have

Paris

• ⟨ Paris ⟩ instance of (P31) ⟨ French city ⟩
  (1st level/second level linked by instance of)
• ⟨ Paris ⟩ instance of (P31) ⟨ Administrative entity ⟩

The second because

⟨ French city ⟩ subclass of (P279) ⟨ Administrative entity ⟩

(second level/ second level linked by subclass of)

French city

• ⟨ French city ⟩ subclass of (P279) ⟨ Administrative entity ⟩
• ⟨ French city ⟩ subclass of (P279) ⟨ French Administrative entity ⟩

⟨ French city ⟩ instance of (P31) ⟨ Type of french administrative division ⟩

⟨ French city ⟩ instance of (P31) ⟨ Type of administrative division ⟩

the last one because

⟨ Type of french administrative division ⟩ subclass of (P279) ⟨ Type of administrative division ⟩

)

To list every types of french administrative divisions (French commune, département, région, ...), we just have to query all the instances of Type of french administrative division. @Nikki: My question about type of administrative division occurs here : to list every items Type of X administrative division, do we need a 4th column to the table ?

If you are not conviced of why we would need a 3rd column, please read Help:Classification#Why_having_items_who_are_both_class_and_instances. TomT0m ^(talk) 09:32, 6 July 2015 (UTC)

I believe I understand now. Yes, having an item in the third collumn would be useful. The problem encountered here is that the item in the third collumn was mistaken for an item in the second collumn by most of the people using it, resulting in the contradictions that normally result from trying to use a single item as both a class and metaclass. I support User:Nikki's suggestion to split this into 2 items, one for each meaning, and each with the correct labels. Popcorndude (talk) 14:37, 6 July 2015 (UTC)

Yes, and document that on the project frontpage. I'll create the type item (Not sure I did not do this once or twice already) ... Happy if we sorted out this finally :) TomT0m ^(talk) 17:59, 6 July 2015 (UTC)

I don't quite understand about which 2 notions you are saying. We already have administrative territorial entity of more than one country (Q15646667) and administrative territorial entity type (Q15617994). --Infovarius (talk) 15:05, 7 July 2015 (UTC)

@Infovarius: For me this item was an item such that Type of French division or type of Australian division to be a instance of : type of administrative division of a country. With an afterthought I'm not sure it belongs to the third column, maybe actually it's a fourth if it's really instance of who is needed. TomT0m ^(talk) 18:56, 7 July 2015 (UTC)

The item should collect all "type of ... division" and actually I don't see a difference if we will use subclass of (P279) or instance of (P31). --Infovarius (talk) 11:39, 9 July 2015 (UTC)

@Infovarius: That's how we end up with Paris being a type of administrative division ... It you query all types of administrative division with a query such as CLAIM[31:(TREE[type of administrative division][][279279])], no, you don't want Paris in the results. You want French region or American state, but not Paris or New-York. TomT0m ^(talk) 11:56, 9 July 2015 (UTC)

I mean we can have administrative territorial entity of France (Q192498) subclass of (P279) "administrative divisions by country" or administrative territorial entity of France (Q192498) instance of (P31) "type of administrative division" - both don't lead to "Paris is a type...". So yes we should choose the meaning of the item. --Infovarius (talk) 16:34, 10 July 2015 (UTC)

@Infovarius: The first one leads to "Paris is an administrative division by country" per transivity of (sub)class membership :) I think we will agree this does not make any sense. This is because we need a third column, it's an opportunity to sort things correctly without adding much complexity, if at all (no property added, just a matter of reading Help:Classification ... The constraint model is able to handle this. TomT0m ^(talk) 10:39, 11 July 2015 (UTC)

Yes, sounds awkward. Ok, may be it's more logical to have instances of this item. --Infovarius (talk) 22:22, 20 July 2015 (UTC)