Wikidata talk:WikiProject Cities and Towns

From Wikidata
Jump to navigation Jump to search

Modelling cities[edit]

There is no formal data model for cities on Wikidata. One of the inspiring reasons for creating this Wikiproject was the free-for-all classification of cities. Specifically, the usage of classes in instance of (P31) who's qualities can otherwise be expressed using properties. For examples, see Chicago (Q1297), City of Brussels (Q239), and New York City (Q60). This trend of superfluous classification likely started with one city and editors have spread it to the others.

I believe that items should be modeled consistently across entities of the same types and should have 1 or as little instance of (P31) values as possible. How I like to use instance of (P31) when modeling data is finding the one most-explicit/straightforward definition of an entity. In the case of cities, this would be city (Q515). Therefore, if city (Q515) is to be the only value used on cities, the others should be removed. However, we must also consider what city (Q515) means in other languages as well as the other superfluous classes or less-definitive options like human settlement (Q486972). Maybe they have some significance that I'm not aware of and are present on items for a reason. If so, please let me know.

To support a change that removes these classes, here is a list of replacements and reasons for replacement. Feel free to add to this list.

and a lot more are obvious as well.

Let me know your thoughts! Lectrician1 (talk) 05:00, 6 December 2021 (UTC)[reply]

I Don't thing its always good idea. There are many local views to this problematic.
Look to Prague (Q1085). There is 15x instance of (P31) and probably only one or two of them can be replaced, some of them are also used for generating lists (eg. Q62049, 5119). Some of them should be moved to some separate property Administrative hiearchy which does not exists.
big city (Q1549591) is not always >100000. In Czech Republic (Q213) are some major cities with ~95000 inhabitants, but are considered as big city (Q1549591) in local sources. And some countries have no big city (Q1549591) from global view, but their capital or big cites are from local view big city (Q1549591).
And from Czech scope - word "město" means both city (Q515) and town (Q3957), so we use separate property municipality with town privileges in the Czech Republic (Q15978299) - and czech is surely not only language with this complication. JAn Dudík (talk) 10:02, 6 December 2021 (UTC)[reply]
Also beware that what is a city, is different in different countries. In Denmark the concept "city" as an administrative unit doesn't exist at all. There are urban areas used for statistical purposes but these are not part of any administrative divisions. The municipalities can contain any number of urban and rural areas, and an urban area can be divided between several municipalities. --Dipsacus fullonum (talk) 16:24, 6 December 2021 (UTC)[reply]
After a discussion with Wikidatians on Telegram, I'm okay with keeping official administrative instance definitions for cities. And maybe they could be used instead of city (Q515) since city (Q515) is quite arbitrary whereas a governmental designation is specific...
As for big city (Q1549591), isn't that just subjective? Sources may call it a "big city", but we're working with data and we should stick to facts to differentiate items and avoid nonspecific definitions. Therefore, we should not use it in the instance.
Looking at Prague (Q1085) a lot of instances seem to intersect each other in definition. For example, municipality with authorized municipal office, administrative district of Czech municipality with authorized municipal office, and municipality of the Czech Republic. These could possibly be reduced, but I don't know exactly how since I do not know how the municipal system works in the Czech Republic. Lectrician1 (talk) 17:21, 6 December 2021 (UTC)[reply]
Czech repulic consists of 14 regions. Every region have 1 (Prague) or more Districts. Every district have 1-n AMO which have 1-x municipalities. Only Prague is munucipality and AMO and District and Region, all in one. It may look strange in case of Prague, but have sense in whole country.
Definition of big city (Q1549591) depends on country. Why 101k yes and 99k no? In global measure might be some definition, but locally is always some specific. JAn Dudík (talk) 21:33, 7 December 2021 (UTC)[reply]
The problem with modelling cities as only instances of city (Q515) is that it is then hard for users to determine what happened with high-electron-mobility transistor (Q1617706) unless there is a method provided for retrieving their instances. Peter F. Patel-Schneider (talk) 20:59, 7 December 2023 (UTC)[reply]
@Peter F. Patel-Schneider did you mean to mention a different item then high-electron-mobility transistor (Q1617706)? Lectrician1 (talk) 15:31, 8 December 2023 (UTC)[reply]
@Lectrician1 Indeed. I meant million city (Q1637706) Peter F. Patel-Schneider (talk) 15:33, 8 December 2023 (UTC)[reply]
@Peter F. Patel-Schneider What do you mean its hard for users to determine "what happened with Q1637706"? Do you mean like when they became cities with populations over 1 million? Lectrician1 (talk) 15:37, 8 December 2023 (UTC)[reply]
@Lectrician1 Your proposal appears to require that there be no instance of (P31) links to million city (Q1637706). But how is a user supposed to treat the (empty) {Q|1637706}}. Is the user supposed to take away that there are no cities that should belong to million city (Q1637706) in Wikidata? My view is that it would be better to have something readily available on million city (Q1637706) that says why million city (Q1637706) and tells the user how to retrieve Wikidata entities that should be instances of million city (Q1637706). Even better would be to actually respond to queries asking for instances of million city (Q1637706) with the appropriate answers. Peter F. Patel-Schneider (talk) 15:43, 8 December 2023 (UTC)[reply]
it would be better to have something readily available on million city (Q1637706) that says why million city (Q1637706) and tells the user how to retrieve Wikidata entities that should be instances of million city (Q1637706)
@Peter F. Patel-Schneider Agreed, we can add that.
Even better would be to actually respond to queries asking for instances of million city (Q1637706) with the appropriate answers.
That's just not possible with our current query engine. It's expected that queryers research the data models of the items they are looking for anyways and they should know not to expect instances of million city. Lectrician1 (talk) 16:12, 8 December 2023 (UTC)[reply]
@Lectrician1 Yes that is how things currently work. I view the requirement that users investigate the classes they query a big problem with Wikidata. Peter F. Patel-Schneider (talk) 17:42, 8 December 2023 (UTC)[reply]

I'm now leaning towards using government designations of cities for their instances instead of city (Q515). city (Q515) is subjective and very lacking in a specific definition, especially across languages. Using government designations on the other hand is exactly what most cities are and would make much more sense for usage as the instance. Lectrician1 (talk) 17:25, 6 December 2021 (UTC)[reply]

At the moment the situation is very messy. However, due to vast regional differences, it is difficult to establish uniform rules, while retaining the automatic creation of query based cross-country lists like w:cs:Seznam největších měst v Evropské unii. But I agree that some instances substituting properties can be quietly deleted (like million city (Q1637706)). --Jklamo (talk) 17:43, 8 December 2021 (UTC)[reply]

Quetly? This is an important designation (at least in Russia) which gives some rights and privileges. Queries are harder to implement that a simple class. --Infovarius (talk) 13:19, 13 December 2021 (UTC)[reply]
@Infovarius What's it's designation in Russia? This designation should be mentioned in the item description... Lectrician1 (talk) 13:54, 13 December 2021 (UTC)[reply]
FYI. Constraints which seem to be the result of this discussion have been added to instance of (P31) in January 2022. --Nw520 (talk)

Locations in general?[edit]

Is there a wikiproject that describes modeling for geo-location data in general? I think there are a lot of issues associated with this that could be better defined... ArthurPSmith (talk) 17:59, 6 December 2021 (UTC)[reply]

@ArthurPSmith I don't think so. I couldn't find any when making the Wikiproject. Lectrician1 (talk) 18:35, 6 December 2021 (UTC)[reply]

Metropolis[edit]

How should we handle big sprawling conurbations like LA, London, or Sydney? If you go to the top of a hill or tall building, and look out over w:en:Greater Sydney (Q3130),[metro 1] you will see a sea of houses, with some high-rise pimples at the central business districts (CBDs, Q738570) of Sydney proper, Parramatta, Chatswood, etc. Other regional centres within the metropolis have less high-rise, like Blacktown or Bankstown. When we say “I'm going to the city” we mean Sydney CBD, not Parramatta. Parramatta is also considered a city, but not the city. Outsiders will say “I'm going to Sydney”, but residents don't. Living in, say Blacktown, we consider ourselves to be already in Sydney. The CBD area we think of as the city spreads over Sydney (suburb), Haymarket, parts of Pyrmont (also suburbs); and is a part of City of Sydney (local government area, LGA), which is part of Sydney (metropolis). The term “Greater Sydney” has come into use in recent decades to distinguish “Sydney” sensu lato from the various “Sydney”s sensibus strictis. I'd love to hear about what terminology people use in other parts of the world.

Unlike Prague, which is said above to also be a municipality and administrative division, Sydney (metropolis) contains multiple cities, some 30ish? LGAs, and many hundreds of suburbs.

The administrative area “City of Blacktown” Q30919 has a population over 300k, which would make it a big city if it were isolated and not part of the larger metropolis. Despite having “city” in its name, the Q-item is correctly instance of (P31)local government area of New South Wales (Q55558200). It contains towns like Blacktown (Q4216448) (suburb, SSC10397) and Seven Hills (Q5680232), residential suburbs like Bungarribee (Q4997625) and commercial suburbs like Huntingwood (Q5945586). Suburb and locality boundaries are officially designated by government, and don't always coincide with postal, electoral, or local-government boundaries. Some suburbs were traditional towns before housing grew up between them, but the bodies that propose and approve the official locality names don't distinguish that. So it's probably appropriate that the Wikidata items are all instance of (P31)suburb (Q188509). Perhaps we should have an item for the historical Seven Hills, NSWinstance of (P31)human settlement (Q486972) distinct from Seven Hills (Q5680232)instance of (P31)suburb (Q188509)? But now I'm digressing, this is meant to be Wikiproject Cities, not Wikiproject Towns.

. ⁓ Pelagic ( messages ) 21:18, 10 December 2021 (UTC)[reply]

Oh, ha! Whilst I was writing this post, the page got renamed to Wikiproject Cities and Towns. ⁓ Pelagic ( messages ) 21:43, 10 December 2021 (UTC)[reply]
@Pelagic
Sydney (Q3130) parallels to London (Q84). It's an arbitrary region that doesn't have an actual government. The Wikipedia articles just discuss it as a general area that has many actual cities inside of it. For these types of items, I'd propose making the primary instance metropolis (Q200250) and not city (Q515).
As for the other definitions of "Sydney" you think are possible, feel free to create items and I'll review them.
City of Blacktown (Q30919) looks fine.
Blacktown (Q4216448) looks like it should be a instance of (P31) of suburb and locality (Q33112019) or suburb/locality (Q7632426) as those are the official government designations (also maybe you should look into making an item for a locality? idk how the australian system works but do whatever you need to do to make it right). If you want to make suburb and locality (Q33112019) a subclass of (P279) suburb (Q188509), go ahead.
Perhaps we should have an item for the historical Seven Hills, NSW?
I don't know how to model the evolution of items. I don't think we have a data model for that in Wikidata yet, but it's something I want to explore in-general. Make the item if you can find evidence of it/references. Otherwise don't. Lectrician1 (talk) 02:48, 14 December 2021 (UTC)[reply]


  1. ↑ Note that w:en:Greater Sydney is a redirect to w:en:Sydney, which has a hatnote “This article is about the Australian metropolis.”

CBD vs gazetted localities[edit]

Continuing the Sydney example, you could argue that the core central business district (Q738570) is the high-rise area bounded by Circular Quay (Q1092910) / Cahill Expressway (Q1025770) (N), Sussex Street (Q7649491) (W), Elizabeth Street (Q5363574) (E) and ? Goulburn Street (Q5588249) (S). And the wider “CBD” business and entertainment destination areas extend into other suburbs like Haymarket (Q2919185), the Darling Harbour redevelopment (the on-land precinct not the watery bay) in Pyrmont (Q2916104), and Barangaroo.

We have Q1852577 P4093 SSC13715. Currently Sydney Central Business District (Q1852577) is an instance of both central business district (Q738570) and suburb (Q188509). If you go to ABS Maps and show the 2016 State Suburbs (SSC) boundaries, you'll see that the suburb is not spatially coincident with any reasonable definition of the CBD.

I'm tempted to go and split Sydney (suburb) from Sydney Central Business District. This is potentially an issue for interwiki links, as the w:en article covers both topics, and articles in other languages could have varying scopes.

For most cities and towns, the “CBD”, “downtown”, or “high-street” will be a small area. It's a useful concept, but do we need a Q-item for the CBD of every single town? What can we state about them other than their extent? Internationally-famous cities, sure, there may be a Wikipedia article, like in Sydney's case. I'm not familiar with Wikivoyage practices, is there a structural need from that side?

This is also highlights the general issue of splitting entities versus maintaining multiple P31s on consolidated entities. ⁓ Pelagic ( messages ) 23:02, 10 December 2021 (UTC)[reply]

Oops, in recent suburb boundaries (at least back to 2011), Darling harbour is in Sydney, not Pyrmont. ⁓ Pelagic ( messages ) 00:03, 11 December 2021 (UTC)[reply]
If a CBD can be governmentally or even societally defined (pop culture) by sources, then go ahead and make an item for it. For example, the CBD for my home city functions well as this:
Chicago Loop (Q70626): central business district of Chicago, Illinois, United States
This is potentially an issue for interwiki links, as the w:en article covers both topics, and articles in other languages could have varying scopes.
Well that means we can't touch it for now. Every day I see new conflicts between Wikidata and Wikipedia concepts covered and soon I'm going to start a proposal about having dedicated Wikidata items for Wikipedia articles. Until we have that, don't touch anything unless you can touch without breaking something or someone possibly breaking something in the future (like merging).
but do we need a Q-item for the CBD of every single town? What can we state about them other than their extent? Internationally-famous cities, sure, there may be a Wikipedia article, like in Sydney's case. I'm not familiar with Wikivoyage practices, is there a structural need from that side?
Like I said, if it's notable and you got sources, you can make an item.
This is also highlights the general issue of splitting entities versus maintaining multiple P31s on consolidated entities.
Ya that's what we're trying to solve with this Wikiproject and like a million other things on Wikidata :P Lectrician1 (talk) 02:58, 14 December 2021 (UTC)[reply]

A little thing: new property[edit]

What do you think about a new property: the category of honorary citizens of the town?

Example:

...etc. Palotabarát (talk) 20:46, 13 December 2021 (UTC)[reply]

@Palotabarát It's used on Wikimedia projects and is a relationship that I can't think of to express in any other way, so yes, a property is needed! Lectrician1 (talk) 02:50, 14 December 2021 (UTC)[reply]
OK, thank's, I started a discussion here, thank you for your support. Palotabarát (talk) 22:32, 16 December 2021 (UTC)[reply]

How to model climate emergency declarations[edit]

@Syddii: Hello members of WikiProject Cities and Towns. I recently noticed climate emergency declaration area in Australia (Q104880945) which is used as superclass for several local government areas. Obviously, ontology-wise using this class as value for instance of (P31) doesn't really make sense (Gawler (Q1847617) -instance of→ climate emergency declaration area in Australia (Q104880945) -subclass of→ climate emergency declaration (Q63438000) is just wrong). So how can this be fixed? Nw520 (talk) 03:16, 28 June 2023 (UTC)[reply]

Note that if a proper modelling solution will be found then at least https://www.wikidata.org/w/index.php?title=Q1533526&diff=prev&oldid=1924930432 https://www.wikidata.org/w/index.php?title=Q1066185&diff=prev&oldid=1924930988 https://www.wikidata.org/w/index.php?title=Q947334&diff=prev&oldid=1924931078 should be undone and redone properly (I removed some also in past) Mateusz Konieczny (talk) 08:25, 28 June 2023 (UTC)[reply]
I've changed the labels and statements on climate emergency declaration area in Australia (Q104880945) to make it refer to the class of areas, not the class of declarations. All statements using the item are places using it as a value of P31, so this change has no knock-on effects. Swpb (talk) 19:53, 6 July 2023 (UTC)[reply]
Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata_talk:WikiProject_Cities_and_Towns&oldid=2026420461"