Commons talk:Categories

From Wikimedia Commons, the free media repository
Jump to: navigation, search


Diffusion of Category:Aircraft by registration[edit]

This meta category has amassed 77,326 subcategories and needs diffusion. Part of the problem is that for quite some time there have been several subcategories in Category:Aircraft by registration by type following the scheme of "Boeing 747 by registration" and the like, but people still keep adding both Category:Aircraft by registration and "Category:<Aircraft type> by registration" to the main categories where individual aircraft by registration are gathered. E.g. Category:EC-MLD (aircraft) is categorised both in "Category:Aircraft by registration" and "Category:Airbus A321 by registration". Contrary to COM:OVERCAT this seems to be the rule at aircraft categories rather than the exception. I am presenting this issue here because Ardfern suggested that it be discussed with Commons:WikiProject Aviation only. However, I don't think that local consensus can trump a Commons-wide policy, so exceptions need to be approved here. De728631 (talk) 16:45, 2 April 2018 (UTC)

Why does a function (set) with 2 parameters - "Aircraft (by registration, by type)" have a superfunction (superset) with 1 parameter - "Aircraft (by registration)"?
To display all parents click on the "▶":

--Fractaler (talk) 09:35, 3 April 2018 (UTC)

That is how Wikimedia categories work. We define more specific categories the further we go down the category tree, and that means that more parameters come into play while the definition set out in a simple top category still remains valid for all elements further down the hierarchy. "Aircraft by registration" is for images where just the registration number is known. "Aircraft by registration by type" is a container for aircraft categories where the registration and the type is known, and "Category:Kawasaki C-1 by registration‎" and the like would be the next level. The problem, however, is that subcategories should only be part of one category level further up the direct line, and not be sorted into two related parent categories. De728631 (talk) 21:07, 3 April 2018 (UTC)
Having so many entries isn't always a reason to diffuse a category. This isn't a standard-type category. Another example of this type of category is Category:People by name, which has even more entries: 366,781 when I checked just now. There are categories that are subsets of that one, such as Category:Men by name and Category:Women by name (see Category:People categories by name for others), but the contents of those categories are also in Category:People by name. We should handle similar categories such as this one the same way. --Auntof6 (talk) 04:36, 4 April 2018 (UTC)
First, what is a "category tree"? If it is a taxonomy , then we have: ROOT <- 1) SUBROOT1 (by A); 2) SUBROOT2 (by B); 3) SUBROOT3 (by A, by B). Examples: "Aircraft by parameters" <- 1) "Aircraft by registration"; 2) "Aircraft by type"; 3) "Aircraft by registration by type" --Fractaler (talk) 08:28, 4 April 2018 (UTC)
Don't place an item into a category and its parent. For example, a black and white photo of the Eiffel Tower should be placed in Black and white photographs of the Eiffel Tower. It should not be placed in both that category and the Paris category at the same time.
Maybe I didn't make it clear enough when I started this discussion, but my main concern is not so much the way we may want to diffuse this category in the future but a massive case of overcategorisation right now. Contrary to the Commons policy on categories, there are probably hundreds of subcategories that are placed into a category and its parent. So my approach is to remove all those subcategories from Category:Aircraft by registration that have already been sorted into a category "by registration by type". This is the commonly accepted standard to solve the issue, but it has been challenged in this case and needs discussion. De728631 (talk) 09:50, 4 April 2018 (UTC)
We have now, for example, 3 sets: 1) Category:Aircraft by registration, 2) Category:Aircraft by type, 3) Category:Aircraft by registration by type (the same for Category:People by name, Category:People by gender, Category:People by name by gender, etc.)‎. So, category tree (by the commonly accepted standard) must be ...? --Fractaler (talk) 10:38, 4 April 2018 (UTC)

┌─────────────────────────────────┘
Isn't this obvious? The category tree should be:

De728631 (talk) 12:11, 4 April 2018 (UTC)
Category:Aircraft by registration, however, may very well contain registration categories like "Category:D-ECAB" if the aircraft type is unknown. Once the type becomes known, the registration category should be placed into "<Aircraft type> by registration" instead. De728631 (talk) 12:15, 4 April 2018 (UTC)
Level 4 - you are right, here COM:OVERCAT. But also I mean (level 2->level 1), why the set Category:Aircraft by registration by type must be a subset of the set Category:Aircraft by type (or Category:Aircraft by registration)
To display the taxonomy below click on the "▶":

?--Fractaler (talk) 13:58, 4 April 2018 (UTC)

For the record, we agree on COM:OVERCAT. As to your question: It is the logical taxonomy for breaking down Category:Aircraft by registration and Category:Aircraft by type. Category:Airbus A380 by registration, Category:Jetstream 31 by registration etc. need to have parent categories and it would be improper to put them directly into Category:Aircraft by registration and Category:Aircraft by type because there are dozens of these "by registration by type" categories. A meta category for hosting them is not only justified but needed to make things more accessible, so that is how the connection between Level 2 and Level 1 works. De728631 (talk) 15:09, 4 April 2018 (UTC)

I'm going to ignore the application of set theory above (IMO it isn't an appropriate model for Commons), as abstract theory is unlikely to be informative to a specific problem.

I will stick to practical concerns. Say I have a photo of the plane with registration G-BOAC. I don't have a clue what sort of plane that is, but if I create its category I can place it in Category:Aircraft by registration based on what I do know. Alternatively, imagine I am seeking images of G-BOAC. I know its registration, so its reasonable to use Category:Aircraft by registration to try to locate it. If its directly in that category, I can find it. If its buried in a "by type" subcategory I cannot find it, as I do not have that information. In both cases, having the individual plane's category in Category:Aircraft by registration is helpful. Removing it from that category is harmful.

To put this a different way, "I want a plane with registration G-BOAC" is not sensibly narrowed down by instead saying "I want a Concorde with registration G-BOAC". In contrast "I want a Concorde" is sensibly refined with "I want a Concorde with registration G-BOAC". That suggests Category:Aircraft by registration by type should be a subcat of "by type" but not "by registration".--Nilfanion (talk) 20:39, 4 April 2018 (UTC)

If you're looking for a specific category G-BOAC, your first start should be the search field anyway. It will guide you directly to the desired category without you having to browse the category tree. It is the fastest solution for "I want a plane with registration G-BOAC", so a direct entry in Category:Aircraft by registration is therefore not even necessary. Also, Category:Aircraft by registration by type includes the "registration" element, so the question would still arise why it is not linked back to Category:Aircraft by registration. Per our category policy, "each category should itself be in more general categories, forming a hierarchical structure." The hierarchical structure would be broken if Category:Aircraft by registration was not involved. Pinging @Joshbaumgartner: who created "by registration by type" as he might want to comment here too. De728631 (talk) 23:24, 4 April 2018 (UTC)
PS: What I'm trying to demonstrate is that navigation in the category realm works both ways, not just top-down. So if I want to browse back from G-BOAC via "Concorde by registration" and further up the tree, I should be able to arrive at "Aircraft by registration" as well. De728631 (talk) 23:31, 4 April 2018 (UTC)
With bottom-up navigation, you can get to Aircraft by registration by some obvious logical route, no matter how its categorised. That is not true for top-down navigation unless it is directly in by registration. Breaking registrations down by type is simply NOT helpful for navigation. Outright deletion of by registration by type is preferable to have it messing up the utility of the by regisration category.--Nilfanion (talk) 16:39, 5 April 2018 (UTC)
Now you are contradicting yourself. A few paragraphs further up, you suggested that "Category:Aircraft by registration by type should be a subcat of 'by type'" rather than by registration while you are now outright opposed to "Breaking registrations down by type"? De728631 (talk) 18:30, 5 April 2018 (UTC)
Uhh.. '"by registration by type" should be a subcat of "by type" rather than "by registration"' is consistent with 'don't break registrations down by type'? The latter statement is stronger, but doesn't contradict the former. If you already know the registration, adding in the type of aircraft doesn't narrow things down further, you already have a unique plane. (As an aside, to me "aircraft type" implies things like "helicopter" or "wide-body airliner" not "Boeing 777"). What benefit is there to any user in removing categories like Category:G-BOAC (aircraft) from Category:Aircraft by registration? IMO the only logical subcats for aircraft by registration are for the countries of registration. That would link all G registered planes together, and would allow G-BOAC to have a sortkey starting with B instead of G - making it slightly easier to find in the still huge list.--Nilfanion (talk) 19:16, 5 April 2018 (UTC)
We do have country-specific categories. Category:Aircraft registered in the United Kingdom is a parent for all G- registration categories, and there are lots of other such categories for more or less any registration prefix. And "type" is the official ICAO designation for what may otherwise be called an aircraft model. Using "model" for general aircraft categories is problematic though, because it should only be used for categories of scale models. Hence the "by type" wording of the subcategories that was rightfully introduced by Uli Elch. De728631 (talk) 19:47, 5 April 2018 (UTC)
And those are the only ones that logically belong under by registration. As they are aspects of aircraft registration, not an otherwise unrelated aspect of aircraft.--Nilfanion (talk) 20:06, 5 April 2018 (UTC)
Are you sure?
Aircraft by registration
`-- Aircraft registered in the United Kingdom
`-- Aircraft registered in France
`---G-BOAC
`---F-IBEX
That way you would empty "Aircraft by registration" of all registration categories, because per COM:OVERCAT they would have to be sorted into the relevant country-specific subcategories, leaving you again with no direct search options. At the moment, "Aircraft by registration" and "Aircraft by registration country" are at the same level in Category:Aircraft registrations and that is a good structure. De728631 (talk) 20:37, 5 April 2018 (UTC)
And I agree with that structure. My point there is if we don't want to merge those two related concepts (the registration code and the registration country), why would we want to link two entirely unrelated categories?--Nilfanion (talk) 21:23, 5 April 2018 (UTC)
I do not understand what "specific problem"? Where can user place Category:G-BOAC (aircraft) based on what user do know or how can user find Category:G-BOAC (aircraft)? Who is the taxonomy for, who is the end user? What is the problem: creating a taxonomy or navigate (by navigator!) through it? Also, just for clarification: set theory is not a model. --Fractaler (talk) 08:42, 5 April 2018 (UTC)
Specific as in actually discussing the particular concern raised. Not discussing general points which could equally apply to any category. The application of set theory to Commons categories is the problematic case. Its based on the assumption that subategories must be subsets. That's clearly not true in many cases.--Nilfanion (talk) 19:16, 5 April 2018 (UTC)
Erm, if a subcategory is not a subset of its parent categories, where is the navigational benefit? Categories in a category tree shall "reflect a hierarchy of concepts, from the most generic one down to the very specific". De728631 (talk) 19:47, 5 April 2018 (UTC)
See this discussion. The navigational benefit is from linking two related concepts, but that relationship is not necessarily that between a set and its subset. The photos of a building in a city are a subset of the photos of the city. The photos of a building built by an architect are not a subset of the photos of the architect.--Nilfanion (talk) 20:06, 5 April 2018 (UTC)
"See this discussion." TLDR, and too much set theory. Still, there is a relationship between the architect and his buildings, so the photos of buildings are a subset of images related to the architect. De728631 (talk) 20:37, 5 April 2018 (UTC)
The short version is that the real issues start to appear at the 2nd order. The building could easily be a subcat of an entirely different city (the birthplace of the architect). That relationship is tenuous, but the two steps to get there are perfectly valid. Its conceivable that someone would place a photo of the building directly in the architect's category; its implausible that they would place it in their birthplace's category. That relationship is clearly not a strict subset-of-subset relationship, in contrast to building-city-country which would be.--Nilfanion (talk) 21:23, 5 April 2018 (UTC)
"Its based on the assumption that subategories must be subsets": first, its based on the assertion that must be a definition ("by list" or "by giving a property"). So, still no definition "by list" or "by giving a property". Also here, " The photos of a building in a city" (Category:Buildings by city? Category:Photos of buildings in a city?) - where can we read the definition of this term? When there are no definitions, then there are disputes. Do we need disputes? --Fractaler (talk) 08:13, 6 April 2018 (UTC)
Set theory is nice, but should not trump what works best for a real application on Commons. Josh (talk) 21:33, 8 April 2018 (UTC)
What does "works best for a real application on Commons" mean? As can be seen from the template with the taxonomy above, for example, in Category:B-6140 (aircraft) -> Category:Aircraft by registration -> Category:Aircraft registrations -> Category:Aviation data -> Category:Data, set theory is simply not used ("B-6140 (aircraft)" is not "Data"). --Fractaler (talk) 07:07, 9 April 2018 (UTC)

We should retain current use of Category:Aircraft by registration. It is an index of all aircraft registrations, regardless of further sub-categorization that can occur. Sub-categorization can be done by type, by country of registration, or by any number of other criteria. It is best if a registration is accurately categorized by all relevant methods, not just one. However, none of that changes the fact that it is both valuable and without harm to have an index that retains a link to all registrations. Since it does no harm and provides value, the current structure and method of using Category:Aircraft by registration should be retained. Josh (talk) 21:33, 8 April 2018 (UTC)

Category:Aircraft by type must be a subcategory of Category:Aircraft by registration? --Fractaler (talk) 07:07, 9 April 2018 (UTC)
Now that would be ridiculous. De728631 (talk) 12:21, 9 April 2018 (UTC)

┌─────────────────────────────────┘
FYI, I have opened a CFD on this, so that people who follow category discussions will see it. --Auntof6 (talk) 16:00, 9 April 2018 (UTC)

Category:Aircraft by registration[edit]

This meta category has amassed 77,326 subcategories and needs diffusion. Part of the problem is that for quite some time there have been several subcategories in Category:Aircraft by registration by type following the scheme of "Boeing 747 by registration" and the like, but people still keep adding both Category:Aircraft by registration and "Category:<Aircraft type> by registration" to the main categories where individual aircraft by registration are gathered. E.g. Category:EC-MLD (aircraft) is categorised both in "Category:Aircraft by registration" and "Category:Airbus A321 by registration". Contrary to COM:OVERCAT this seems to be the rule at aircraft categories rather than the exception. I am presenting this issue here because Ardfern suggested that it be discussed with Commons:WikiProject Aviation only. However, I don't think that local consensus can trump a Commons-wide policy, so exceptions need to be approved here. De728631 (talk) 16:45, 2 April 2018 (UTC)

Why does a function (set) with 2 parameters - "Aircraft (by registration, by type)" have a superfunction (superset) with 1 parameter - "Aircraft (by registration)"?
To display all parents click on the "▶":

--Fractaler (talk) 09:35, 3 April 2018 (UTC)

That is how Wikimedia categories work. We define more specific categories the further we go down the category tree, and that means that more parameters come into play while the definition set out in a simple top category still remains valid for all elements further down the hierarchy. "Aircraft by registration" is for images where just the registration number is known. "Aircraft by registration by type" is a container for aircraft categories where the registration and the type is known, and "Category:Kawasaki C-1 by registration‎" and the like would be the next level. The problem, however, is that subcategories should only be part of one category level further up the direct line, and not be sorted into two related parent categories. De728631 (talk) 21:07, 3 April 2018 (UTC)
Having so many entries isn't always a reason to diffuse a category. This isn't a standard-type category. Another example of this type of category is Category:People by name, which has even more entries: 366,781 when I checked just now. There are categories that are subsets of that one, such as Category:Men by name and Category:Women by name (see Category:People categories by name for others), but the contents of those categories are also in Category:People by name. We should handle similar categories such as this one the same way. --Auntof6 (talk) 04:36, 4 April 2018 (UTC)
First, what is a "category tree"? If it is a taxonomy , then we have: ROOT <- 1) SUBROOT1 (by A); 2) SUBROOT2 (by B); 3) SUBROOT3 (by A, by B). Examples: "Aircraft by parameters" <- 1) "Aircraft by registration"; 2) "Aircraft by type"; 3) "Aircraft by registration by type" --Fractaler (talk) 08:28, 4 April 2018 (UTC)
Don't place an item into a category and its parent. For example, a black and white photo of the Eiffel Tower should be placed in Black and white photographs of the Eiffel Tower. It should not be placed in both that category and the Paris category at the same time.
Maybe I didn't make it clear enough when I started this discussion, but my main concern is not so much the way we may want to diffuse this category in the future but a massive case of overcategorisation right now. Contrary to the Commons policy on categories, there are probably hundreds of subcategories that are placed into a category and its parent. So my approach is to remove all those subcategories from Category:Aircraft by registration that have already been sorted into a category "by registration by type". This is the commonly accepted standard to solve the issue, but it has been challenged in this case and needs discussion. De728631 (talk) 09:50, 4 April 2018 (UTC)
We have now, for example, 3 sets: 1) Category:Aircraft by registration, 2) Category:Aircraft by type, 3) Category:Aircraft by registration by type (the same for Category:People by name, Category:People by gender, Category:People by name by gender, etc.)‎. So, category tree (by the commonly accepted standard) must be ...? --Fractaler (talk) 10:38, 4 April 2018 (UTC)

┌─────────────────────────────────┘
Isn't this obvious? The category tree should be:

De728631 (talk) 12:11, 4 April 2018 (UTC)
Category:Aircraft by registration, however, may very well contain registration categories like "Category:D-ECAB" if the aircraft type is unknown. Once the type becomes known, the registration category should be placed into "<Aircraft type> by registration" instead. De728631 (talk) 12:15, 4 April 2018 (UTC)
Level 4 - you are right, here COM:OVERCAT. But also I mean (level 2->level 1), why the set Category:Aircraft by registration by type must be a subset of the set Category:Aircraft by type (or Category:Aircraft by registration)
To display the taxonomy below click on the "▶":

?--Fractaler (talk) 13:58, 4 April 2018 (UTC)

For the record, we agree on COM:OVERCAT. As to your question: It is the logical taxonomy for breaking down Category:Aircraft by registration and Category:Aircraft by type. Category:Airbus A380 by registration, Category:Jetstream 31 by registration etc. need to have parent categories and it would be improper to put them directly into Category:Aircraft by registration and Category:Aircraft by type because there are dozens of these "by registration by type" categories. A meta category for hosting them is not only justified but needed to make things more accessible, so that is how the connection between Level 2 and Level 1 works. De728631 (talk) 15:09, 4 April 2018 (UTC)

I'm going to ignore the application of set theory above (IMO it isn't an appropriate model for Commons), as abstract theory is unlikely to be informative to a specific problem.

I will stick to practical concerns. Say I have a photo of the plane with registration G-BOAC. I don't have a clue what sort of plane that is, but if I create its category I can place it in Category:Aircraft by registration based on what I do know. Alternatively, imagine I am seeking images of G-BOAC. I know its registration, so its reasonable to use Category:Aircraft by registration to try to locate it. If its directly in that category, I can find it. If its buried in a "by type" subcategory I cannot find it, as I do not have that information. In both cases, having the individual plane's category in Category:Aircraft by registration is helpful. Removing it from that category is harmful.

To put this a different way, "I want a plane with registration G-BOAC" is not sensibly narrowed down by instead saying "I want a Concorde with registration G-BOAC". In contrast "I want a Concorde" is sensibly refined with "I want a Concorde with registration G-BOAC". That suggests Category:Aircraft by registration by type should be a subcat of "by type" but not "by registration".--Nilfanion (talk) 20:39, 4 April 2018 (UTC)

If you're looking for a specific category G-BOAC, your first start should be the search field anyway. It will guide you directly to the desired category without you having to browse the category tree. It is the fastest solution for "I want a plane with registration G-BOAC", so a direct entry in Category:Aircraft by registration is therefore not even necessary. Also, Category:Aircraft by registration by type includes the "registration" element, so the question would still arise why it is not linked back to Category:Aircraft by registration. Per our category policy, "each category should itself be in more general categories, forming a hierarchical structure." The hierarchical structure would be broken if Category:Aircraft by registration was not involved. Pinging @Joshbaumgartner: who created "by registration by type" as he might want to comment here too. De728631 (talk) 23:24, 4 April 2018 (UTC)
PS: What I'm trying to demonstrate is that navigation in the category realm works both ways, not just top-down. So if I want to browse back from G-BOAC via "Concorde by registration" and further up the tree, I should be able to arrive at "Aircraft by registration" as well. De728631 (talk) 23:31, 4 April 2018 (UTC)
With bottom-up navigation, you can get to Aircraft by registration by some obvious logical route, no matter how its categorised. That is not true for top-down navigation unless it is directly in by registration. Breaking registrations down by type is simply NOT helpful for navigation. Outright deletion of by registration by type is preferable to have it messing up the utility of the by regisration category.--Nilfanion (talk) 16:39, 5 April 2018 (UTC)
Now you are contradicting yourself. A few paragraphs further up, you suggested that "Category:Aircraft by registration by type should be a subcat of 'by type'" rather than by registration while you are now outright opposed to "Breaking registrations down by type"? De728631 (talk) 18:30, 5 April 2018 (UTC)
Uhh.. '"by registration by type" should be a subcat of "by type" rather than "by registration"' is consistent with 'don't break registrations down by type'? The latter statement is stronger, but doesn't contradict the former. If you already know the registration, adding in the type of aircraft doesn't narrow things down further, you already have a unique plane. (As an aside, to me "aircraft type" implies things like "helicopter" or "wide-body airliner" not "Boeing 777"). What benefit is there to any user in removing categories like Category:G-BOAC (aircraft) from Category:Aircraft by registration? IMO the only logical subcats for aircraft by registration are for the countries of registration. That would link all G registered planes together, and would allow G-BOAC to have a sortkey starting with B instead of G - making it slightly easier to find in the still huge list.--Nilfanion (talk) 19:16, 5 April 2018 (UTC)
We do have country-specific categories. Category:Aircraft registered in the United Kingdom is a parent for all G- registration categories, and there are lots of other such categories for more or less any registration prefix. And "type" is the official ICAO designation for what may otherwise be called an aircraft model. Using "model" for general aircraft categories is problematic though, because it should only be used for categories of scale models. Hence the "by type" wording of the subcategories that was rightfully introduced by Uli Elch. De728631 (talk) 19:47, 5 April 2018 (UTC)
And those are the only ones that logically belong under by registration. As they are aspects of aircraft registration, not an otherwise unrelated aspect of aircraft.--Nilfanion (talk) 20:06, 5 April 2018 (UTC)
Are you sure?
Aircraft by registration
`-- Aircraft registered in the United Kingdom
`-- Aircraft registered in France
`---G-BOAC
`---F-IBEX
That way you would empty "Aircraft by registration" of all registration categories, because per COM:OVERCAT they would have to be sorted into the relevant country-specific subcategories, leaving you again with no direct search options. At the moment, "Aircraft by registration" and "Aircraft by registration country" are at the same level in Category:Aircraft registrations and that is a good structure. De728631 (talk) 20:37, 5 April 2018 (UTC)
And I agree with that structure. My point there is if we don't want to merge those two related concepts (the registration code and the registration country), why would we want to link two entirely unrelated categories?--Nilfanion (talk) 21:23, 5 April 2018 (UTC)
I do not understand what "specific problem"? Where can user place Category:G-BOAC (aircraft) based on what user do know or how can user find Category:G-BOAC (aircraft)? Who is the taxonomy for, who is the end user? What is the problem: creating a taxonomy or navigate (by navigator!) through it? Also, just for clarification: set theory is not a model. --Fractaler (talk) 08:42, 5 April 2018 (UTC)
Specific as in actually discussing the particular concern raised. Not discussing general points which could equally apply to any category. The application of set theory to Commons categories is the problematic case. Its based on the assumption that subategories must be subsets. That's clearly not true in many cases.--Nilfanion (talk) 19:16, 5 April 2018 (UTC)
Erm, if a subcategory is not a subset of its parent categories, where is the navigational benefit? Categories in a category tree shall "reflect a hierarchy of concepts, from the most generic one down to the very specific". De728631 (talk) 19:47, 5 April 2018 (UTC)
See this discussion. The navigational benefit is from linking two related concepts, but that relationship is not necessarily that between a set and its subset. The photos of a building in a city are a subset of the photos of the city. The photos of a building built by an architect are not a subset of the photos of the architect.--Nilfanion (talk) 20:06, 5 April 2018 (UTC)
"See this discussion." TLDR, and too much set theory. Still, there is a relationship between the architect and his buildings, so the photos of buildings are a subset of images related to the architect. De728631 (talk) 20:37, 5 April 2018 (UTC)
The short version is that the real issues start to appear at the 2nd order. The building could easily be a subcat of an entirely different city (the birthplace of the architect). That relationship is tenuous, but the two steps to get there are perfectly valid. Its conceivable that someone would place a photo of the building directly in the architect's category; its implausible that they would place it in their birthplace's category. That relationship is clearly not a strict subset-of-subset relationship, in contrast to building-city-country which would be.--Nilfanion (talk) 21:23, 5 April 2018 (UTC)
"Its based on the assumption that subategories must be subsets": first, its based on the assertion that must be a definition ("by list" or "by giving a property"). So, still no definition "by list" or "by giving a property". Also here, " The photos of a building in a city" (Category:Buildings by city? Category:Photos of buildings in a city?) - where can we read the definition of this term? When there are no definitions, then there are disputes. Do we need disputes? --Fractaler (talk) 08:13, 6 April 2018 (UTC)
Set theory is nice, but should not trump what works best for a real application on Commons. Josh (talk) 21:33, 8 April 2018 (UTC)
What does "works best for a real application on Commons" mean? As can be seen from the template with the taxonomy above, for example, in Category:B-6140 (aircraft) -> Category:Aircraft by registration -> Category:Aircraft registrations -> Category:Aviation data -> Category:Data, set theory is simply not used ("B-6140 (aircraft)" is not "Data"). --Fractaler (talk) 07:07, 9 April 2018 (UTC)
@Fractaler:: What is your proposal then? Which of the links you listed is invalid and should be broken? Josh (talk) 22:19, 9 April 2018 (UTC)

We should retain current use of Category:Aircraft by registration. It is an index of all aircraft registrations, regardless of further sub-categorization that can occur. Sub-categorization can be done by type, by country of registration, or by any number of other criteria. It is best if a registration is accurately categorized by all relevant methods, not just one. However, none of that changes the fact that it is both valuable and without harm to have an index that retains a link to all registrations. Since it does no harm and provides value, the current structure and method of using Category:Aircraft by registration should be retained. Josh (talk) 21:33, 8 April 2018 (UTC)

Category:Aircraft by type must be a subcategory of Category:Aircraft by registration? --Fractaler (talk) 07:07, 9 April 2018 (UTC)
Now that would be ridiculous. De728631 (talk) 12:21, 9 April 2018 (UTC)
@De728631:: It is ridiculous. Category:Aircraft by type is NOT a subcategory of Category:Aircraft by registration, nor should it become one, nor is anyone proposing that. As I stated above, we should retain current use of Category:Aircraft by registration. Josh (talk) 22:19, 9 April 2018 (UTC)

┌─────────────────────────────────┘
FYI, I have opened a CFD on this, so that people who follow category discussions will see it. --Auntof6 (talk) 16:00, 9 April 2018 (UTC)

Please see discussion at Commons talk:Categories#Diffusion of Category:Aircraft by registration. Auntof6 (talk) 15:51, 9 April 2018 (UTC)

Moved discussion text to this page so it will reflect in real time on both Commons talk:Categories and Commons:Categories for discussion. Josh (talk) 21:45, 9 April 2018 (UTC)
  • There is no problem with having all single-aircraft registration categories in the main "Aircraft by registration" acting as a super-category. This is not uncommon practice. A standardisation of the "by type" subcategories is always a good thing, of course. Huntster (t @ c) 19:25, 9 April 2018 (UTC)

There are some valid reasons to rethink exactly how Category:Aircraft registrations is structured. Never mind the hashing about whether or not a guideline is being obeyed or whether we are properly applying set theory, none of that is terribly valuable. The category does however beg some more clarity and streamlining. There are a couple issues which we can deal with in pieces, or as a whole. Josh (talk) 23:22, 9 April 2018 (UTC)

1 - xxxx (aircraft) categories are aircraft registrations, not aircraft. However, they are often treated as aircraft, especially since they say 'aircraft' parenthetically. This is not a problem for most common usage, but is exposed in corner cases and when analyzing the category structure. Keep in mind an aircraft may be assigned several registrations over its life, and some registrations may be assigned to different aircraft over time. Specific sub-categories of an aircraft registration category can be created to show its application to different aircraft (e.g. Category:N305FA (aircraft) into Category:N305FA (Boeing 737) and Category:N305FA (MD-83)). Proper names should be 'Aircraft registration N305FA' with sub-cats 'Aircraft registration N305FA assigned to Boeing 737 c/n 28662' and 'Aircraft registration N305FA assigned to MD-83 c/n 49398'. I am not proposing renaming these categories, unless someone is up for moving 75,000+ categories. The current abbreviated names are fine, but we should have a better description of what exactly those categories cover.
2 - Category:Aircraft by registration is named incorrectly. As noted above, the sub-cats are aircraft registrations, not aircraft, so the correct title should be Category:Aircraft registrations (flat list) or some other such appropriate title to indicate it is an index of all aircraft registrations ordered alpha-numerically. As it is, the current name adds to the confusion referred to in note 1 above. It may be appropriate to make this category a hidden cat while we are at it. Once this is done, sub and meta cats can be moved directly under Category:Aircraft registrations.
3 - Military identification numbers are not consistently treated. These are sometimes treated as aircraft registrations and other times as serial numbers or some other unrelated tree. Category:Aircraft registrations should cover all individual aircraft identifications assigned by authorities, military or civil. Sub-categorization can break down between assigning authorities for those that it is helpful for, but not all users will know what the issuing authority is for a particular identifier. No rename is needed, but a better description is required to make it clear what the category covers.
4 - Category:Aircraft by registration country is named incorrectly. As above, a more clear and concise name should be used, such as Category:Aircraft registrations by country of issue, to make it clear that the items within are aircraft registrations and that they are ordered by the country which issued the registration. It should be listed directly under Category:Aircraft registrations and not under Category:Aircraft by registration/Category:Aircraft registrations (flat list). Category:Aircraft by registration continent should get similar treatment, though 'continents' do not issue registrations, countries do.
5 - Category:Aircraft by registration by manufacturer and type are incorrect. They should be renamed Category:Aircraft registrations by aircraft assigned and sub-cats of that can parallel the categorization of Category:Aircraft to the level appropriate. Category:Aircraft registrations by aircraft assigned should be directly under Category:Aircraft registrations.

Some tweaks like these would allow the continued use of aircraft registration categories essentially as they have been used for the 75,000+ registrations in place, while at the same time adding clarity and cleaning up the structure of the category quite a bit. They will hopefully go some way to satisfying concerns over COM:OVERCAT and the set theory issues raised by De728631 (talk · contribs) and Fractaler (talk · contribs). Josh (talk) 23:22, 9 April 2018 (UTC)

"What is your proposal then?": set theory requires a definition, and therefore here, in the disputed case, it makes sense to give definitions to the term. What definition should the term "B-6140 (aircraft)" have for a more general term to be the term data"? The same for "aircraft by registration", "aircraft registrations", etc. --Fractaler (talk) 07:22, 10 April 2018 (UTC)
@Fractaler: As stated in the list above, definition of Category:B-6140 (aircraft) is an 'aircraft registration'. Not sure what definition you are looking for beyond that. Josh (talk) 16:46, 10 April 2018 (UTC)
I don't think we need to differentiate between xxxx (aircraft) and xxxx (aircraft registration) and all subsequent namings. Apart from Category:Temporary aircraft registrations that are used for test and transfer flights, registrations are seldom changed over the life of an aircraft frame and the registration is therefore often synonymous with the single airframe it got assigned to. We already have Category:Re-used aircraft registrations and its appropriate sub-categories as you showed above.
@De728631: You are incorrect that registrations are seldom changed; it is common practice to change a commercial aircraft registration several times during its life, especially when it changes ownership. I would not advise eliminating the existing sub-categorization of xxxx (aircraft) into xxxx (specific aircraft) categories. Assuming synonymy between an aircraft and its registration is a mistake. As stated, I am not proposing that these categories be renamed, but merely that we have better definition of them as being specifically related to that aircraft registration. Josh (talk) 16:46, 10 April 2018 (UTC)
Well, you wrote "Category:Aircraft by registration is named incorrectly. ... the correct title should be Category:Aircraft registrations (flat list) or some other such appropriate title", or "Category:Aircraft by registration by manufacturer and type are incorrect. They should be renamed Category:Aircraft registrations by aircraft assigned". Isn't that renaming? Apart from Category:Aircraft registrations (flat list), I think this is unnecessary, and imho Category:Aircraft registrations by aircraft assigned would be outright confusing. Btw, you created the two latter categories (by registration by manufacturer / by registration by type [model]) last year, so how come you changed your mind now? As I see it, the focus is already on the registration numbers now – even with names like "xxxx (aircraft)". If it's really that common for commercial registrations to be changed, Category:Re-used aircraft registrations with xxxx (specific aircraft) subcategories should become more populated though. Different aircraft should not be lumped into a single registration category. De728631 (talk)
My apologies for not being clear. I don't propose changing the xxxx (aircraft) naming scheme. I do however, think that the meta cats they are in should be renamed per my suggestions above. You are right that some of them are ones I created myself under flawed names. I named them as I did in order to keep with the naming of Category:Aircraft by registration, but I wasn't thrilled by it at the time, and I am even less so now. I'm not sure what you are concerned about with lumping. As it stands now, if a registration is applied to multiple aircraft (which is less common than one aircraft having multiple registrations), then it should be broken down (see Category:N305FA (aircraft)). The main registration category should be also categorized in Category:Re-used aircraft registrations. That is current practice, and I don't think anyone is suggesting changing it. Josh (talk) 21:19, 10 April 2018 (UTC)
Military ID numbers are a problem though. Apparently there are in fact two major approaches among the armed forces of how to apply such registrations, namely using an aircraft's generic serial number (e.g. US Air Force, Italy) or issuing an unrelated ID (Germany, UK, Netherlands, etc.) Sometimes like in Italy or Spain, there are even two parallel schemes of markings on a single aircraft, such as an internal squadron ID (e.g. 41-12) and a permanent serial number. This has already led to inconsistent category schemes as in Category:Military aircraft registered in Spain or Category:Military aircraft registered in Italy (see the MM##### serials). So these need some consistency. De728631 (talk) 09:54, 10 April 2018 (UTC)
The US is no different than Germany or the UK: none use a 'generic' serial number, but instead assign their own numbers per whatever system they have established at the time. Some of these systems adopt the serials already assigned by other agencies or the manufacturer, but again, the sytems are set by each individual issuing authority. What is fundamentally different are identifiers that are assigned for the service life of an aircraft (such as the US Navy's BuNo) vs. those that are assigned to indicate organizational assignment and may be changed throughout its service life (such as the US Navy's tactical codes). However, in all cases, just as with civil registrations, the categories are for the identifier, not the airframe, and thus they should all be handled within the same consistent structure regardless of local differences in how such numbers are devised. Josh (talk) 16:46, 10 April 2018 (UTC)
Thanks for clarifying the service-life ID vs tactical code schemes. It is essentially what I tried to write above but maybe it didn't come through. I agree that in the future we should not use any tactical codes for "registration" categories but stick to BuNos, serials and other such official "top-level" IDs. Where applicable, we should redirect existing "tactical" categories to categories with the official registration number, e.g. Category:43-28 (aircraft)Category:UD.13-28 (aircraft). De728631 (talk) 20:48, 10 April 2018 (UTC)
I totally agree with you that the 'tactical codes' and the like should not be necessarily considered aircraft registrations, while 'serial numbers' like BuNos, etc. should be under aircraft registrations. I also agree that it is curently not consistent and has been hard to know exactly how to proceed with those kinds of categories. We can have 'tactical code' categories, but they should be kept in their own category. The difficulty will be that many users may not be aware of the differences. Josh (talk) 21:19, 10 April 2018 (UTC)
definition of Category:B-6140 (aircraft) is an 'aircraft registration' : now Category:B-6140 (aircraft) does not have any definition. But, for example, Category:Civil aircraft by country, Category:Airliners of Spain, Category:Four-engine airliners, Category:China Southern Airlines have (even human readable, not to mention the machine-readable, as, for example, in Wikidata or Commons' Category:Airbus A380). --Fractaler (talk) 12:17, 11 April 2018 (UTC)
Do you mean that the definition should be reflected in the name of the category? De728631 (talk) 15:09, 11 April 2018 (UTC)
Now on the page Category:B-6140 (aircraft) we can see such static information: Airbus A380-841, cn/serial number: 120, *China Southern Airlines 2013 to date as B-6140. No "is an 'aircraft registration'" on the page. And the pages from the examples have definitions on their pages ("China Southern Airlines is an airline based in Guangzhou in the Guangdong province of the People's Republic of China (PRC)", etc.). --Fractaler (talk) 06:26, 12 April 2018 (UTC)
Or, for example, Wikidata's definitions:

--Fractaler (talk) 13:11, 12 April 2018 (UTC)

Those infoboxes are well suited to gallery pages, but not so much for categories. Wikidata doesn't have items for individual aircraft registrations as far as I am aware. I just looked it up and there are no items with instance of: Q838849 (aircraft registration) Josh (talk) 21:17, 12 April 2018 (UTC)
Do you mean these items? About definition: in order to be able to display the definition (by version of WD, if there is no version of Commons) on a category page, I'm now trying to make a template {{DescriptionWD}} (using Module:Wikidata description). For example, "aircraft registration": registration and identification assigned to an individual aircraft by national aviation authorities --Fractaler (talk) 07:59, 13 April 2018 (UTC)
Try Template:Individual aircraft and Template:Wdd. Josh (talk) 19:07, 13 April 2018 (UTC)
@Fractaler: Yes, none of those items you linked are instances of aircraft registrations. Josh (talk) 19:18, 13 April 2018 (UTC)

What should the order of category names in a file be?[edit]

The question is, which order should the categories be in, if, for example, a photo contains multiple items. In this example file of backs of buses in Tallinn, Estonia, the categories are currently in this order at the bottom of the file.

As people start reading from left to right, the date/year/time and location categories first help establish context for people who are from outside Estonia, and then would move on to the details, as they keep on reading:

  • July 2007 in Estonia
  • 2007 in Tallinn
  • Buses in Estonia photographed in 2007
  • Tallinna Autobussikoondis
  • TAK buses in green and white livery
  • Articulated buses in Estonia
  • Volvo buses in Estonia
  • Volvo B10MA
  • Aabenraa buses in Estonia
  • Buses built in 1984
  • 1984 Volvo buses
  • Ajokki buses in Estonia
  • Ajokki City N
  • Buses built in 1985
  • 1985 Volvo buses
  • Scania buses in Estonia
  • Scania CN113ALB
  • 1990 Scania buses

There appears to be no exact policy with regard to the order in which categories are placed in a file. Some categories don't have much order, because they are added/moved with automated or semi-automated tools, which I have also made good use of.

Neither does policy define the order of category names in a file, but current instructions are shown in such way, that gives preference to main objects, but not the overall context, that I think should be discoverable first.

If I have a potential dispute with an IP user about this, then who should have the upper hand? -Mardus /talk 11:00, 5 April 2018 (UTC)

There is no preference to the order. Some people prefer a thematic structure, some prefer alphabetical. Others (the vast majority) just use the somewhat random order you get from the automated tools. You wouldn't have the "upper hand" over an IP, and that dispute should be handled like any other content dispute - through discussion and consensus building. If edit warring occurs, it might need admin intervention. But IMO its not worth fighting over.--Nilfanion (talk) 19:22, 5 April 2018 (UTC)
"category names in a file" is "machine-readable data". To access an object according to a taxonomy, the horizontal order of categories (the order of categories indicated on the file page) does not matter. --Fractaler (talk) 09:28, 6 April 2018 (UTC)
I agree that it generally isn't worth arguing as there isn't a paricular order. However, I don't think that it doesn't matter at all since readers actually do use categories to browse similar images, categories aren't used only for queries and such. Generally, key feature of an image is its subject (what it depicts), at least on a project like Commons, where images are supposed to have educational value rather than artistic value, and as per this page this is the first priority when choosing categories for an image. This allows an assumption that readers are also most interested in subject when they are looking for similar images, and these categories should be the easiest to find, somewhere at the top or at least in the first half of the list of categories. So, if particular user systematically puts most important categories at the latter part of long category lists, like here, and also fills first half with categories like "date in some location" that didn't even exist until recently (probably because they are least useful), then this really doesn't seem constructive.
I don't quite get it what do "details" mean above, or if there is something that could be considered a non-detail if the main subject is considered a "detail". There also seems to be some confusion about the purpose of categories if someone tries to use them to "establish context", rather than simply tying similar images. I doubt if any reader would work out that category list is place where context lies. Generally, description is used for this purpose. If anything, and if description is insufficent, then categories may occasionally reveal what the subject is. 90.191.81.65 09:26, 7 April 2018 (UTC)
"about the purpose of categories": can categories have a purpose? A person can have a goal and uses tools (such as, for example, taxonomy) to achieve his goals. machine-readable data (in this case) means that the reader can get any order of categories that he needs: alphabetically, by the date of adding categories, by the number of supercategories of supercategories, by the length of the name, in order according to some user, by the least path to the root, etc. Ie, wars for the order in which the categories should go, become meaningless. --Fractaler (talk) 12:59, 7 April 2018 (UTC)
Well, we categorize images rather than leave them uncategorized. Why so, I'd say, is the purpose of categories. This page, to some extent, covers the purpose too. As said, I understand categories can be used for queries and other stuff that relates to machine-readabilty, but people also make use of category list at the footer of file page. This list to date is fixed to order given in syntax of given page. That said, even though we generally don't expect certain order in this list, and we necessarily can't avoid random order, then we still can avoid notably peculiar order introduced by particular user.
So far, there are aspects of browsing the site which often make users add categories in order which isn't entirely random, they find some order more useful than other. So category order isn't meaningless, and I find it inevitable that category order should be discussed sometimes, hopefully rather seldom than often, but still. Though, I concur, category order should become meaningless. Hopefully it becomes more so when structured data comes to Commons. So that perhaps certain categorizing schemes could be dropped, especially everything related to dates which combined into categories is technically just silly. 90.191.81.65 14:05, 7 April 2018 (UTC)
"Hopefully it becomes more so when structured data comes to Commons": yes, and it looks like everything is coming to this. --Fractaler (talk) 17:09, 7 April 2018 (UTC)
  • Whatever is most useful to the reader. So place those at the front which the reader of this object might wish to navigate to. Either because they're an eponymous category and this is a gallery page, Wikipedia lead article or similar. Or else those which are primary defining categories for the object. The categories of "Tuesdays in Bohemia" and "Taken with a camera phone" should be low down, if present at all. Andy Dingley (talk) 16:30, 7 April 2018 (UTC)