Wikipedia talk:Categorization

From Wikipedia, the free encyclopedia

[edit] the listing of category page links at bottom of article pages

Should be alphabetical. To try to list them according to supposed importance is utterly WP:POV. And I would like to see alphabetic listing of cat page links get mention in WP categorization guidelines for I have encountered users who list according to a supposed list of importance. There are however certain exceptions that are being met with, such as Category:Presidents of the United States links - clearly that this link is of primary importance for those who have held this office is universally accepted Mayumashu (talk) 17:58, 11 March 2008 (UTC)

Alphabetizing categories has be discussed several times and rejected several times. Check the archives. -- ☑ SamuelWantman 23:45, 11 March 2008 (UTC)

It's hard to believe, given the large number of comments, but a careful check of the archives in fact turns up not a single talk entry on the matter - maybe the talk occurred elsewhere. It would be nice to see what the argument(s) against the idea have been given how clearly evident attempting to judge a cat links's importance is POV (and that if a majority of users deem a link unimportant (enough) to have here, that it can be deleted after discussion at WP:Cats for discussion) Mayumashu (talk) 04:32, 12 March 2008 (UTC)

Wikipedia editing in general is often about selecting and arranging information according to its importance. Opinions may differ, but that doesn't mean common sense and consensus can't be applied to reach a satisfactory result. This applies to the ordering of categories as much as to anything else.--Kotniski (talk) 10:34, 12 March 2008 (UTC)

I know that there have been discussions, because I took part in many of them. Here are the mentions I found (there may be more):

--☑ SamuelWantman 20:38, 12 March 2008 (UTC)

If I'm not mistaken, there was also a discussion at WP:AWB about ordering. Consensus was to not alphabetize the categories when using AWB. Personally, I agree with Mayumashu. Who decides importance? Alphabetical isn't POV. --Kbdank71 20:43, 12 March 2008 (UTC)

Editorial decision-making about importance, weight and order are part of editing every page. This is no different with categorization. Using the same logic, we could decide to order the sections of every article alphabetically. -- ☑ SamuelWantman 20:49, 12 March 2008 (UTC)

Really grateful for you listing those links. You make a good point in that editing, as it occurs at various levels, is inherently based on POV and therefore POV can never be entirely done away with. I think though as well that most users desire too convention (I know I do). One idea (that involves a little more complexity) would be to have category pages graded on an importance scale (we could have, say, three levels) that would be maintained through discussion/nomination to change. (Then within one scale one could alphabetize the members of that set.) Assigned levels to begin with could be changed through nomination. This would systemize the editorial decision-making while establishing a compromise between mechanical alphabetic listing that ignores common sense and editing based solely on POV. (True however, isn t it that the ones that most people would care about are likely the ones most people would agree on) Mayumashu (talk) 00:10, 18 March 2008 (UTC)

(undent) have category pages graded on an importance scale - Given all the things that need to be done to improve Wikipedia, is this really, really necessary? -- John Broughton (♫♫) 13:56, 21 March 2008 (UTC)

I personally prefer it as going 0-9; A-Z as it looks much neather than if they were in any old order. D.M.N. (talk) 20:06, 28 April 2008 (UTC)

Wouldn't it make more sense to have similar categories near to each other? Like nationality or ethnicity categories together? That would greatly ease navigation. Besides, I'm not sure how any cat listing can be considered neat. It's a string of phrases. The neatest you could get is to arrange them by length. -Freekee (talk) 02:30, 2 May 2008 (UTC)

Personally, when it comes to categories, there are few things that annoy me more than purely alphabetical lists. My approach is to group them thematically, putting those that deal with occupation and/or avocation toward the top, with life dates, etc. at the bottom. It's definitely an art, not a "science" with rigid rules. Cgingold (talk) 12:38, 3 June 2008 (UTC)

[edit] Eponymous categories (topic categories)

This has no doubt been discussed many times before (including a short distance above on this same page), but we need to do something about the way we handle these "eponymous" categories (basically the same as the "Topic Categories" of Sam's proposal). I can't see any really satisfactory consistent way of dealing with them without asking for a software improvement. Basically the issue is which categories the topic category ought to be placed in, and which the corresponding article should be placed in. I notice that different policies on this are used in different parts of Wikipedia, and none of them give totally satisfactory results (either categories are unnecessarily duplicated, or else either articles or topic categories are missing from their natural categories).

So what I propose asking the developers to consider is something like this: define a way of mapping an article to a corresponding topic category (e.g. by some variation on the [[Category:...|...]] syntax). On the topic category page, as well as its own parent categories, show (separately) the categories of the mapped article. Meanwhile in all category pages, on the list of member articles, next to the name of any article which is mapped to a topic category, include a button linking to that category. And at the same time, maybe introduce a more general syntax for placing a category in another category as a member, not as a subcategory (causing it to be listed among the member articles, not among the subcategories).

The policy would then be to place articles in their natural categories, and topic categories only in categories of categories. Everything would show up in the appropriate place, and information would not have to be entered in duplicate to achieve this. We would also have the ability to distinguish in general between member categories and subcategories (which would be useful in the case of categories of categories, although that's a separate issue).

Thoughts?--Kotniski (talk) 09:16, 28 March 2008 (UTC)

After thinking about categories for over 3 years now, my thought is that they need an overhaul. The problem is, and has been, that there is not a consensus about what categories are, what functions they should have and what functions they shouldn't have. Perhaps, what is needed is some discussion about the big picture. Should categories be a system of classification? a system of indexing? a system for browsing? a way to do database searches? It seems to me that many of the people who spend time categorizing are more concerned with classification (where articles belong) than with browsing (where articles should be found). Since this is a wiki, we have let things evolve on their own, and this has led to people working at cross purposes. On top of this, the wiki software has not been friendly to categorization work -- there is no history and things cannot be easily reverted. The current system could work sufficiently if we could all agree that categories should be optimized for classification, indexing, browsing or searching, but it cannot be optimized to handle all these goals. I have stopped actively creating and populating categories because of the frustration wrought by these insufficiencies.

So perhaps an entire new system needs to be designed. If so, we should be clear on what we want it to achieve and create clear criteria that any such system needs to fulfill. It might not be Categories that fulfill all the criteria we create, but something different or a combination of features. -- ☑ SamuelWantman 06:26, 29 March 2008 (UTC)

And there's not even a consensus that there need to be a consensus on any of these things. I share your analysis in terms of dissatisfaction with the amazingly ad hoc status quo, and the desirability of something much more systematic. However, the realities of the wiki process, and in particular the tendency of "topic ownership" to win out over any sort of global standards and consistency, makes me more than a little skeptical as whether it's an area that one is likely to get much traction in, compared to the considerable effort involved. Alai (talk) 06:48, 29 March 2008 (UTC)

I'm not actually so totally pessimistic (at least, not yet). Agreed, it can sometimes be hard to get anything changed on Wikipedia, but on other occasions it turns out to be surprisingly easy (like with the hidden/maintenance categories thing, which I never expected to happen so fast and with so little community resistance). I think that if a few more software improvements were introduced (like the one I'm proposing here, and like the intersecting categories thing, if work is still being done on that) then categories could be made to work in a useful and fairly consistent way. Not necessarily optimized for any single goal, but working well for a number of purposes. --Kotniski (talk) 09:28, 29 March 2008 (UTC)

Some features would be helpful, but it seems that for anything more than "tagging" you really need to be able to keep a history and be able to watch and revert changes. Imagine what articles would look like without histories, watch pages and reverts. Considering that these are missing from categories it is quite amazing that they are in as good a shape as they are. I don't think they can be much improved without adding those key "wiki" features. Perhaps, as an alternative, we could start to systematically define and approve categories and their taxonomies with the stated goal that all undefined and unapproved categories would be deleted after a certain date, and afterwards all category taxonomies would need to be "preapproved". I proposed something like this at one point and it was not received well. I'm not trying to be pessimistic, what I'm pointing out is that the status quo is about as good as it can be considering the technical limitations. If we want to move past those limitations, we should have a very broad discussion about our goals. The technology that runs this wiki is very much responsible for its strengths and weaknesses. The decisions that are made and the design of any upgrades will have huge impact on the policies and practices that result. In a sense I am saying that the we need to put the cart in front of the horse and steer it in the right direction. -- ☑ SamuelWantman 01:13, 30 March 2008 (UTC)

"Meanwhile in all category pages, on the list of member articles, next to the name of any article which is mapped to a topic category, include a button linking to that category." - amazingly, I suggested something like this recently (before reading this). See here: "ie. Have a softlink [to the eponymous category] appear automatically next to the article name." I think this would solve the "problem" of eponymous categories overnight, though there would still be the problem of where to categorise them. Carcharoth (talk) 15:00, 18 April 2008 (UTC)

Yep, looks like we're on the same wavelength. As for where to categorise the eponymous categories, if my suggestion were adopted, then they would only need to be placed in appropriate "categories of categories", such as Category:People categories or subcategories thereof. For that purpose it would help if it were possible to add a category to another category as a member rather than as a subcategory, but when I proposed that on Bugzilla it was quickly shot down by a developer.

I think we should discuss and develop here a comprehensive proposal about category-related software enhancements to take to the developers, based on our various ideas. Maybe we could even get a developer or two involved in the dialogue (or maybe there's a better forum for such purposes?) It's unfortunate that there seems to be this divide between the WP community (in fact between the various language WP communities as well) and the developer community.--Kotniski (talk) 08:45, 19 April 2008 (UTC)

[edit] Live Music Archive category

I would like to propose a category listing all bands who permit their recordings to be placed on the Live Music Archive. Here is a list. What can we call this? Does anyone agree? Tenacious D Fan (talk) 19:44, 3 April 2008 (UTC)

[edit] New category added - opinions please

Hi. I've created a new category - Category:Satellite operators. The rationale for why I thought it was necessary is on the categories' talk page. Before I spend too much time on adding articles to the category etc., can I have some feedback on whether the consensus is that the category is suitable, in need or redefinition, or just plain not needed :-) Cheers. CultureDrone (talk) 19:03, 18 April 2008 (UTC)

To link to a category, use the following mark-up: [[:Category:Satellite operators]], which produces: Category:Satellite operators. The category looks fine to me. Carcharoth (talk) 21:48, 18 April 2008 (UTC)

[edit] Which cat?

I don't know what to categorize List of Elena Paparizou awards as. Grk1011 (talk) 21:40, 23 April 2008 (UTC)

[edit] Standard work

I'd like to see the above turned into a Wiki category. Every known text or study considered to be such should be classified as such on Wikipedia. Thanks for any help. By the way, the above is currently under consideration for deletion as a mere definition. I happen to oppose. Any views on the matter, for or against, will be more than welcome - by me at least. --Ludvikus (talk) 00:48, 25 April 2008 (UTC)

The biggest problem with the article, Standard work is that of original research It needs to have some references. As such it really is not possible to create a category for it. Dbiel ^(Talk) 01:28, 25 April 2008 (UTC)

[edit] Question

Am I right in saying, that in articles at the bottom, categories should be listed as 0 to 9; then going from A to Z? I am pretty sure this has been mentioned somewhere before, but I just wanted clarification. So just as an example, instead of it being:

[[Category:Japan]] [[Category:1989]] [[Cateogry:Animals]] [[Category:Water]]

in an edit box, it should instead look like:

[[Category:1989]] [[Cateogry:Animals]] [[Category:Japan]] [[Category:Water]]

Am I right, or am I talking dribble? :)) D.M.N. (talk) 17:50, 28 April 2008 (UTC)

No, I think it's generally agreed that catetgories should be ordered by importance and logic rather than alphabetically.--Kotniski (talk) 18:48, 28 April 2008 (UTC)

There is a discussion about this higher on the page. --☑ SamuelWantman 19:57, 28 April 2008 (UTC)

Thanks. There needs to be a proper consensus on this though, so people are not left confused. D.M.N. (talk) 20:07, 28 April 2008 (UTC)

[edit] Duplication

The current guideline on whether to include subcategories on articles is majorly at odds with the whole concept of depopulating large parent cats. For example, there will be many geographical features that go beyond one subdivision of a country, say, where each subdivision has a "Geography of Foo" category, so there will be many items at the national level leading one to follow the rote if, then instructions to put subdivision specific articles at the national level and at the subdivision level. This is what is done in the Italian WP for Italian comuni (municipalities) - they appear in the provincial, regional and national categories. Does that make them easier to navigate? Not sure, but if we did likewise in the US, every town would be in its county, state, and along with a few hundred thousand others at the US level. If we believe that specific and narrow categories aid navigation and ponderously large categories hinder it, we should make the default: don't duplicate. I can see the argument made at the subcat page, that a casual reader would not know to look under a subcategory, because he or she didn't know whether so-and-so-actor won an Oscar or not. Most readers access categories first through articles, so if so-and-so-actor were in the Oscar subcat and someone wanted to find non-winners, accessing the Oscar winners cat will show the obvious parent sticking out for the user to find the rest of non-winning actors so duplication saves one click and really doesn't aid navigation. Same for most of the other categories: articles should be put in the narrowest category that applies because navigation in categories with thousands of entries is basically not possible. I think the page ought to be updated to reflect this. Carlossuarez46 (talk) 02:08, 2 May 2008 (UTC)

Isn't this already covered in WP:SUBCAT? --BrownHairedGirl (talk) • (contribs) 03:30, 2 May 2008 (UTC)

This is coverd by WP:SUBCAT. I don't know about other people, but I use categories to search for things that I have forgotten the name of. So small categories can make this task much more difficult. If for example, I don't know the nationality of an actor I am looking for, I won't be able to find him unless I look through dozens of subcategories. What I fail to understand about the objections to large categories is why they bother people if you also have the option to select a smaller subcategory for browsing? Also, see the next section for another reason why we need larger categories... -- ☑ SamuelWantman 07:29, 6 May 2008 (UTC)

[edit] Category intersection update

I recently updated the project page with instructions for using Search to perform category intersections and category unions. (I'm not sure how long this feature has been available in Search, and it makes me wonder how many other undiscovered features there might be!) The current version available by using Search has some major drawbacks;

The results are not displayed like a category page.
The results are not alphabetized.
Articles categorized by using templates (or anything transcluded) will not be found by searching.
Subcategories and articles are displayed intermixed.

I saw (Wikimedia CTO) Brion Vibber this past weekend, and had a good conversation with him about the future of categories. He seemed very clear that a better version of Category Intersection and Union is coming. He thinks it will initially take the form of dynamic page lists, which will be able to post queries on any page. There has already been an extension that does this for quite some, but it has not been implemented on Wikipedia for technical reasons. Brion said that he expects that a new version of dynamic page lists will be made compatible with Wikipedia. This feature will make it possible to create pages that look like categories, using intersections and unions, without having the articles actually be in those categories.

There are basically two ways to make this feature work with categories. We can have very narrowly defined categories (like we have now) and use category unions to get larger groupings, or have very broadly defined categories and use category intersection to get the smaller groupings.

Both Brion and I think that ultimately we will want to have large broad categories. Here are the reasons why. Let's say you are looking for a category intersection for two traits that have not been intersected in the current system, perhaps because the result is not considered a notable defining characteristic, or some other reason considered an overcategorization. For example, we don't have Category:LGBT Buddhist actors. If you want to do this starting with our current categorizations you would first have to create the union of all the articles in the subcategories of Category:LGBT actors and intersect this with the union of all the articles in the subcategories of Category:Buddhists. It is already possible to do these queries, but since most categories have been divided into small pieces, most of the queries that users might want to make are ponderously difficult and time consuming tasks.

On the other hand, if we have broadly defined categories, this query would be fairly simple. It would simply the intersection of Category:LGBT people with Category:Actors and with Category:Buddhists.

It seems clear to me that it will be much much easier to do these queries if we have broad categories to intersect.

From previous discussion, higher on this page, several options were discussed:

We can repopulate bigger categories (I like to call them topic level categories) and delete all the smaller ones that are intersections. For example we'd repopulate Category:LGBT people, Category:Actors and Category:Buddhists and delete Category:LGBT actors
We can repopulate topic level categories and leave the smaller ones as they are. If it becomes difficult to navigate from the topic level categories to the subcategories, the subcategories can be put in a new intermediate subcategory. This already is a common practice for categories with large numbers of subcategories. For example we'd repopulate Category:LGBT people, Category:Actors and Category:Buddhists and keep Category:LGBT actors. If it were difficult to navigate to the subcategories of Actors, we could add Categories:Actors (index) to hold all the subcategories.
We can create new categories for the topic level categories that would be fully populated and make them hidden. In this example we'd create Category:LGBT people (all), Category:Actors (all) and Category:Buddhists (all) which would all be subcategories of the existing categories.
We can create the same new categories mentioned in #3 without making them hidden.
We can rename our existing categories that just contain subcategories and call them indexes, and then recycle the old name for a new subcategory that contains all the articles. For example, Category:LGBT people would be renamed Category:LGBT people (index), Category:Actors would be renamed Category:Actors (index), and Category:Buddhists would be renamed Category:Buddhists (index). A new fully populated Category:Actors would be a subcategory of Category:Actors (index), etc...

If we make the categories hidden, people are much less likely to see and use them. For the purpose of doing queries it would be easiest to just repopulate our existing categories. There is no need to delete any existing categories until we have a better functioning Category Intersection process and created or repopulated broad categories.

The difference between #2, #3, #4 and #5 is basically how categories are named, the order of the hierarchy, and if they are hidden. To illustrate:

OPTION #2

Category:Actors -- repopulated to contain all articles about actors. It has just one subcategory:
- Category:Actors (index) -- just contains subcategories.
  - all the current subcategories of Actors.

OPTION #3

Category:Actors -- just contains subcategories (as it is now).
- all the current subcategories of Actors.
- Category:Actors (all) -- A new category to contain all articles about actors. (Hidden)

OPTION #4

Category:Actors -- just contains subcategories.(as it is now).
- all the current subcategories of Actors.
- Category:Actors (all) -- A new category to contain all articles about actors. (Not Hidden)

OPTION #5

Category:Actors (index) -- just contains subcategories.(Category:Actors as it is now, but renamed).
- all the current subcategories of Actors.
- Category:Actors -- A new category to contain all articles about actors. (Not Hidden)

I'm thinking #5 is the best option. There were several suggestions to do something similar in the previous discussions about this, but by renaming the categories we'd solve some additional problems. Having the large categories simply named would make it easier to use them for queries, and it would also likely be the category name most likely used by inexperienced users. I'd like to hear what other people think. --☑ SamuelWantman 08:29, 7 May 2008 (UTC)

I don't particularly like this. First, how many (hundreds) of categories would become indexes. Where does it stop at Actors and not People? Converting them would require a great deal of work and affect nearly every single article. What I would strongly support is an intercategory search feature such as the one you listed on WP:VPPR or a special page. I understand the subcategory problem, but I'd be willing to wait for it to be effective. I also don't like how some categories would be populated and then have thousands of articles in them making them unnavigable. If I was forced to choose one of the above, it would strongly be #3, the same of yours but order reversed, as every topic category would not need to be renamed; and the category shown would not be the extrememly large one. Reywas92^Talk 21:23, 7 May 2008 (UTC)

I feel like we are stuck in the status quo. With a full implementation of category intersection and union, we can start replacing many of our categories with dynamically created lists. For example, eventually all of the "fooian fooer" categories can be removed and replace with categories for nationality and occupation. This means that we would not be constrained by peoples nationality when looking at occupation and vice versa. As we create more and more categories this becomes more and more of a problem. Please take a look at Wikipedia:Category intersection for more about how we might fully implement a solution. Also, I do not understand why everyone thinks a large category is unnavigable. They all have tables of contents. Even Category:Living people is navigable. If you have the choice of looking through either a big OR small category, you will pick the one most suited for you task. -- ☑ SamuelWantman 07:19, 8 May 2008 (UTC)

I'm guessing that there is a technical reason for this, but why not just allow the search to search through subcategories? - jc37 22:30, 7 May 2008 (UTC)

Yes, the problem is technical. Even with fully populated categories speed and server load will be an issue. This is what has prevented implementation up till now. --☑ SamuelWantman 07:19, 8 May 2008 (UTC)

{Start of discussion moved from Village Pump}

Well I imagine that the category "Bridges in New York City" is in the category "Bridges in New York" which is in the category "Bridges in the United States" (hence the name "sub-category"). So why doesn't someone just make the category search recursive? Kevin Baas^talk 20:37, 7 May 2008 (UTC)

That would work, but would also be incredibly slow. Intersecting "Suspension bridges" with "Bridges in the United States" takes two database searches. Intersecting with "Bridges in the United States" and all subcategories would take over a hundred, and would need to deal sensibly with the incredible mess that the category hierarchy is. --Carnildo (talk) 20:47, 7 May 2008 (UTC)

This is a much-awaited feature and ought to be quite useful for a variety of things - like searching for images of certain objects under a particular license, for example. I don't think the category hierarchy ought to be flattened, but instead a technical solution should be sought for intersecting category hierarchies (based for example on a per-category cache of all articles in its category hierarchy). Dcoetzee 20:53, 7 May 2008 (UTC)

(edit conflict) There shouldn't be any category loops, if that's what you're implying. That would contradict the definition of "category". And even if there were, one could write an algorithm that creates a list of all the loops, and then one that checks for loops when a new category is added to a category. But I highly doubt there would be any. I wouldn't imagine it would be overly deep, or overly wide, for that matter, when compared against the size of wikipedia.

And really, if this is not done then to get the same result "flat" levels would have to be mantained, and it is not feasible to expect this to be done manually, so someone would have to create a mechanism that will add an article to all the super-categories whenever its added to a subcategory. That would move the load from time-of-access to time-of-create, but it would make for some HUGE categories and some articles in lots of categories.

I know it would be slow, but I don't think it would be "incredibly" slow -- category hierarchies aren't really that deep; on average, they're pretty shallow. And I think it might ultimately be necessary, regardless.

The analogy I used to consider it was to directories/folders in a computer file system. Often times one wants to search them recursively, and often times one does. One would never think of duplicating each file link to the parent directories. Kevin Baas^talk 21:04, 7 May 2008 (UTC)

Yes, there shouldn't be any category loops, but there are. Yes, the typical hierarchy isn't deep, but that doesn't mean that nobody will ever try an intersection involving Category:Contents. Category intersection needs to deal with Wikipedia as it exists, not Wikipedia as we wish it would be. --Carnildo (talk) 21:48, 7 May 2008 (UTC)

(ec)This is a useful start. m:User:Duesentrieb/CatScan can do category intersections that go some levels deep, but such searches are - as noted above - significantly slower.

I suppose that as storage continues to get less expensive and CPU cycles continue to get cheaper, one solution would be to create "invisible categories". Then if article A is in category X, and category X has parent categories Y and Z, article A could be "invisibly" in categories Y and Z as well - that is, the database would consider the articles to be the latter two categories, but they wouldn't display on article A's page. [Restricting this sort of "upward" addition of categories to (say) two levels up could reduce the number of articles that end up in large numbers of categories, while still addressing the needs of most readers and editors. And it would avoid the problem of category cycles.) -- John Broughton (♫♫) 21:54, 7 May 2008 (UTC)

{End of discussion copied from Village Pump}

If there are category loops, when way to deal with this is to remove them. A script could be written to compile a list of all the loops (if there are any, which i doubt). And then they can be cut, one at a time. RE: "Yes, the typical hierarchy isn't deep, but that doesn't mean that nobody will ever try an intersection involving Category:Contents" - As/regards categories like Category:Contents - such special categories could be restricted (or be ignored/have no effect). Beyond that, the average case is what matters, not the worst case. And the average case is shallow. Kevin Baas^talk 17:50, 8 May 2008 (UTC)

I moved the discussion above from Village Pump. If any of the participants end up on this page, I hope you will read the longer analysis of the problem at the top of this section, and comment on the options presented. -- ☑ SamuelWantman 07:26, 8 May 2008 (UTC)

OPTION #6

Category:Actors contains, as far as a search is concerned, articles in that category and in subcategories.
This is done by creating an additional field (flag) to show if a category of a page is actual or virtual.
"Actual" means "category link is visible on the page"; "virtual" means "page belongs to this category by virtue of belonging to a subcategory.

Yes, this involves programming. And it requires additional server resources and storage (which are getting cheaper all the time). What it doesn't require is (a) a massive reconfiguration of the existing category structure; (b) the permanent destruction of much of the value of subcategories and (c) retraining the entire Wikipedia community on how categories and subcategories should and should not be used, and then enforcing the change at the cost of both effort and editors who leave the project because they disagree with the rationale for the change. -- John Broughton (♫♫) 15:37, 8 May 2008 (UTC)

OPTION #6 sounds wonderful (and would also populate the category 'Actors (all)' of Option #3). My Qs about SW's options are: do the parent/index categories get populated automatically (immediately or in due course) if something is added lower down; and who decides which are 'index categories'? (Eg we have footballers, subcat of sportspeople.) -- roundhouse0 (talk) 16:07, 8 May 2008 (UTC)

Perhaps I was not clear, or perhaps I'm not understanding your concerns. Except for Option #1 (which I am NOT advocating for a short term solution), no subcategories would be destroyed. So I don't understand your item (b). As for (a), some options (#3 and #4) are not massive "reconfigurations" at all, they just involve adding some more categories. The ones that do involve "reconfigurations" are mostly just renaming. So I don't see that your item (a) is an inevitable result of creating some large categories. And (c) offers a wonderful rationale for creating the second edition of "The Missing Manual"! Seriously, the current system makes more sense for categorizers than for users. Yes, often people are looking for a small set of clearly defined articles. But, I think it is just as likely (I won't venture a guess if it is more or less) that people want to browse through a larger collection of articles. I'm often frustrated that I can't browse through all actors, bridges in the UK, films, novels, film directors, Civil War battles, etc... This insistence of creating microscopic categories is an artifact of the wiki technology left over from when categories were first introduced. I am advocating that we make it possible to browse more than one way. -- ☑ SamuelWantman 20:20, 8 May 2008 (UTC)

It seems that an excellent long-term programming solution for categorization (and many other features) would be to implement Semantic MediaWiki on WP. And ideally as some kind of link between English and other language WPs. (See my vague ideas, written up some time ago, here.)--Kotniski (talk) 16:33, 8 May 2008 (UTC)

It seems like the question was misrepresented to me. Rather than saying "Bridges" is broader than "Bridges in New York City", I would say that "Bridges in New York City" is a complex or compound category, whereas "Bridges" is a simple one. Or that "Bridges in New York City" is an explicit intersection, which could be broken up into "Bridges" + "In New York City" (type + location) I think this would be a good move.

However, this wouldn't solve the problem that a search for "In America" would have to include all the states and cities. The two obvious options are :

recursive search - more processing power, no additional storage

pre-recursed categories - no additional processing power, more storage

But there is also a third option, in-between the two:

recursive search w/result caching.

And as regards the above proposals, (except for #6), it should be kept in mind that category:actors is probably in category:people - i.e. that the proposals would have to apply to every hierarchy level, not just the top, because people aren't always going to be searching just the top, or just the bottom. Kevin Baas^talk 18:04, 8 May 2008 (UTC)

Not necessarily. This only would involve a specific class of category -- roughly the plural noun categories. For example, there would be no change at all to "Film", the change would only effect "Films". It is also not clear if all of the categories within the class would need to be repopulated. For example, I think "Actors" should be fully populated, but I don't think "Entertainers" has to be. For me the distinction is whether we can expect a user to know where to go when they are browsing for something. I'll give you examples. If someone wants to browse through Civil War battles, will they know which campaign to look in? If I go to "Film directors", will they know what nationality to look in? Whenever we break up categories into smaller pieces that are not defining characteristics we do a disservice to Wikipedia readers. -- ☑ SamuelWantman 20:20, 8 May 2008 (UTC)

I would like to point out that the developers are struggling to come up with an implementation that will do a category intersection using fully populated categories. Suggesting alternate solutions which involve even faster servers and complications does not seem very likely to come about. So my question about #6 is how these "virtual" flags are implemented. Is this sort of like the hidden categories? Would each article be put in the larger categories (a) the same way as now by editing the page and adding a link to the category, but with the addition of setting this new flag, or (b) would all the articles in a subcategory be part of a larger one by virtue of setting the flag in the subcategory? Choice (a) isn't all that different from having a separate hidden category (Option 3). It would require more programming to implement and would function much differently for users. It might be harder to understand conceptually for editors. (b) I think this has been called "category flattening" in the past. I'd like to hear what the developers think of this, but ultimately I don't think it is needed (read on).

Conceptually, it might help if we imagine what things would be like in the future. If intersections and unions are implemented, and there are dynamic page lists that create them, then a large fraction of our current categories could be replaced with dynamically created pages. From the point of view of the user, this would look very similar to what we have now, with the exception that many more categories would be fully populated. The proposal outlined at Wikipedia:Category intersection suggests check boxes for selecting the categories that you want to intersect. So if an actor is categorized as "American" "People" "in France" "in Paris" "Actor" "Buddhist", you could select which ever you want by clicking on a check box next to each, and then clicking a button to create a dynamically created list with your selection, for example "Buddhist Americans in Paris". There is no need to keep the intersections listed on the article pages. Category pages for these intersections could be replaced with list pages that are dynamically created each time they are visited. So for users who are browsing through categories, all the current information would remain on the replacement lists. Instead of finding subcategories, the user would find (dynamically created) lists with the same information. There would be more options for those getting to categories from the article pages, as every possible intersection would be available. As we discuss the options we have now, I hope we can keep in mind and discuss where we ultimately want to end up. What we do now should be a step in that direction as well as offering current utility using search. -- ☑ SamuelWantman 20:20, 8 May 2008 (UTC)

It is difficult to visualise some of this, though the examples help. Are there test wikis where people can view the results of what this sort of thing would look like? I've supported category intersection and still do, but would really love to see working examples that can be tested and played around with, rather than mock ups and pre-selected examples that may look good but don't really tell us what to expect. Carcharoth (talk) 01:43, 9 May 2008 (UTC)

I've heard that Wikinews already uses dynamic page lists. I don't know an example page to mention. The interfaces mocked up at Wikipedia:Category intersection have not been created. Flickr does "tag intersection" -- all tags are one word, so if you search for more than one word you get the intersection. I'm hoping that perhaps we can mock up sample category hierarchies using the different options under consideration and see how they look and feel. -- ☑ SamuelWantman 07:03, 9 May 2008 (UTC)

[edit] Option #6

Perhaps an example would be useful. If article A is in category X, and category X has parent categories Y and Z, article A would be "virtually" in categories Y and Z as well - that is, the database would consider the articles to be the latter two categories, but they wouldn't display on article A's page. Nor, looking at category pages Y or Z, would you see article A listed; you'd only see subcategory X listed. For a database viewpoint, there would be a three records:

Article A belongs to category X, flag="actual"
Article A belongs to category Y, flag="virtual"
Article A belongs to category Z, flag="virtual"

For search purposes, the flags are ignored, which accomplishes "flattening" without eliminating subcategories. But when actually displaying a category page, flags are used, as noted above, to hide/ignore "virtual" pages in a category.

This option is different than having an "Actors (all)" category in a number of ways: first, it avoids the confusion of having both an "Actors" and an "Actors (all) category; second, users are searching on actual/real categories, not "(All)" categories; third, no one has to decide at what level of a hierarchy of categories the "(all)" categories will be created.

How would this actually work in practice? Let's say that someone adds category X to article B. Then multiple records would be created: one actual record (Article B belongs to category X) and multiple virtual records (Article B belongs to categories Y, Z, etc.)

If you envision categorization from the perspective of a page in a lowest possible category (there are no daughter/sub-categories), then there could be a large number of categories above it (an inverted tree). To make option #6 feasible, I think, it would be necessary to limit the number of levels that the system goes up in building records of virtual membership in categories - perhaps only 2 or 3 levels up, at least initially. That would make searching on large (upper categories) like "People" ineffective, but that seems to be the case with all of the other options.

Again, the advantage of all this is that it is completely behind the scene - it requires no "flattening", no choice of what level(s) or categories to create "flattened" categories, no new (hidden) categories (just hidden/virtual records/membership), and no changes in editor behavior/training. It's also compatible with any approach that splits up currently manually-created intersection categories ("Amerian architects", which would become "American" and "architect"). -- John Broughton (♫♫) 01:54, 9 May 2008 (UTC)

The big question for me is would this actually work in practice. Let me see if I'm understanding this correctly, All articles put in a child category automatically get added, albeit virtually, to their parent and grandparent (and perhaps great-grandparents). So does that mean that all articles in Category:Federal Reserve economists get added virtually to Category:American economists and Category:Federal Reserve and the parents of those two categories Category:American people by occupation, Category:Economists by nationality, Category:American social scientists, Category:Central banks, Category:Independent agencies of the United States government, Category:Economy of the United States, Category:Banks of the United States, Category:Interest rates, Category:Monetary reform and perhaps even the parents of those nine categories, but for now I'll stop at two levels. Likewise, Category:People from New York City also get categorized virtually in Category:New York City, Category:People by city in New York and Category:People by city in the United States along with the parents of those three categories, Category:Cities in New York, Category:Coastal cities in the United States, Category:New York metropolitan area, Category:Regions of New York, Category:People by city in the United States by state, Category:People from New York, Category:American people, Category:People by city, Category:American people by state, and Category:Categories by city in the United States. Therefore the intersection between Category:Banks of the United States, and Category:New York metropolitan area will include former Federal Reserve chairman Alan Greenspan. BTW, I started from a random article. Option 6 would regularly lead to intersections returning articles that should not belong. This doesn't seem right. -- ☑ SamuelWantman 06:53, 9 May 2008 (UTC)

From what I'm envisioning, the intersection would be something like Category:Banks and Category:In New York City metropolitan area. And since Alan Greenspan is not a bank, he would not be in the intersection. If he was in the banks category, he would be removed from it. Perhaps there would be a category Category:Banking that he would be in. (and that banks would be a subcategory of.) Kevin Baas^talk 00:02, 10 May 2008 (UTC)

I don't think Alan Greenspan should pop up if you intersect Category:Banking and Category:In New York City metropolitan area or if you intersect Category:Banks of the United States, and Category:New York metropolitan area. -- ☑ SamuelWantman 07:33, 10 May 2008 (UTC)

I think there will be bizarre results however this is done as some editors think that, given 2 categories, one must be a parent of the other. Look at the parents of Category:Latin American and Iberian Britons - I expect we can get Geri Halliwell into Banks without much trouble. -- roundhouse0 (talk) 14:06, 10 May 2008 (UTC)

Oh, and for an example from Wikinews of dynamic page lists w/category intersections, just follow any of the portal links at the top of the main page. Kevin Baas^talk

The Alan Greenspan example, above, involves a mistake in the categorization scheme: Category:Banks of the United States is both about banking and about individual banks. There should be a category Category:Banking in the United States; and both Category:Banks of the United States and Category:Federal Reserve should be subcategories of that. If this mistake is fixed, then (ignoring levels for the moment) Alan Greenspan is at the intersection of Category:Unites States banking and Category:New York metropolitan area, which is where one would expect to find a person who lives in that geographical area and is involved in banking (but is not a bank).

More to the point - I had assumed that Option #6 behaved exactly the same way as all the other options with respect to what articles would fall into various intersections, except that some other options involve picking levels for (virtual or other) flattening - say, "Actors". I'd welcome any proof to the contrary, but without that, any examples of odd results from category intersections for Option #6, not present in other options, is simply because those other options would be doing far, far fewer intersections. And perhaps that's a good thing - in which case we can modify Option #6 to say that only categories marked as (say) {{SubcatsForIntersection}} would have virtual records added for articles in subcategories. Or that the other options involve lots and lots of manual decisions as to which subcategories to include.

As for Wikinews, while it's interesting, the dynamic lists being generated seem to look something like this:

category=published
category=Crime and law
category=Africa
count=5

Which isn't quite the same challenge as at Wikipedia, which all the levels of categorization here. -- John Broughton (♫♫) 22:16, 10 May 2008 (UTC)

It may be possible to get #6 to work, but that would likely involve a fair amount of recategorization, even with the addition of your {{SubcatsForIntersection}} template. The problem is that categories and their parents sometimes have a "A is a B" relationship, but often they have a "A is related to B" relationship. In many previous discussions here and elsewhere, the "A is related to B" relationship has been found to be acceptable. It adds to the richness of our categorization system. Now it is possible to clean it up, and I have even supported cleaning it up (by adding manual "See also" links instead of subcategorizing), but I want to point out that along with proposing a system that no developer has expressed an interest in creating, and that does not work with our current search capabilities, it will also need a good amount of categorization work and disruption to become functional. Options 2 through 5 would work with search as it is, would not change the relationships of any current categories, and some options do not even change the articles contained in any existing categories. Dynamic page lists WILL work with all the levels of categorization that we have, as long as they are populated. It is for this reason that I want to have populated categories that we can use with our current search capabilities and with future page list capabilities. On top of that, we'd be making our categories more intuitive. Newbies are always adding people to categories like "Actors" and "American people" and "Biologists", we wouldn't have to fight this natural impulse any more. Ultimately categories should be usable for doing searches, for browsing, for indexing broad and narrow topics as well as functioning as a system of classification. It cannot currently do all these things at the same time. With populated high-level topic categories, and category intersections being created dynamically, it could do all these things. Best of all, it would be easy to explain: Each defining characteristic gets its own category. So my question for John is, what is the advantage of keeping our strangely kludged together system which: would not be easier to browse; would not be easy to explain; adds many intersection categories that would not be needed; and isn't as clear as a system of categorization?

It is possible to implement the repopulating of large topic categories in stages:

STAGE ONE: we create new broad categories such as described in Option #3 and #4. They could start out hidden and named Category:Fooers (all), and at a later stage be made visible if there is a consensus to do so. As hidden categories, they would not be at all disruptive to our current system.
STAGE TWO: At some time in the future, after many of these categories are created, we can have a discussion about renaming them to match option #5. This would make it easier and intuitive to do searches.
STAGE THREE: When category intersection is fully implemented we can discuss deleting all the intersections to complete option #1. Intersection categories could be replaced with dynamic page lists. Hopefully, the user interface will make it easy to generate these lists (see: Wikipedia:Category intersection, and the pages could still be reached by browsing through categories (they would be pages instead of categories). Articles would no longer be cluttered with intersection categories, yet many more intersections could be created.

Each stage in the process would be incremental, and each would be advantageous. If I have learned anything about Wikipedia in the last 4 years, it is that we have to work with what we have, and change has to be incremental, improving things at each step. Stage one and two would be useful now as things are.

I don't want to dismiss Option 6 totally, but if we are to consider it, it needs to be fully fleshed out, and some discussions should happen with Brion and other developers to find out if it has any chance of being implemented. There are probably some daunting technical questions. Try posting on the wiki-tech mailing list, and see what type of response you get. If the developers are positive, so might I. -- ☑ SamuelWantman 07:24, 11 May 2008 (UTC)

Option 6 retains one keep point, that is the ability to easily located mis categorized and under categorized articles. That's a plus going forward as long as most if not all articles are in sub categories. Vegaswikian (talk) 17:18, 21 May 2008 (UTC)

[edit] Subcategorization and the "secondary categorization rule"

Confounding Wikipedia:Categorization_and_subcategories#Secondary_categorization_rule reads:

When an article is put into a subcategory based on an attribute that is not the first thing most people would think of to categorise it, it should be left in the parent category as well.

This includes articles placed in ethnic subcategories within national menus, for example articles in Category:African American baseball players should also be left in Category:American baseball players.

How am I supposed to know if an attribute is "not the first thing most people would think of to categorise it?" Why would it be likely that someone would think of Hank Aaron as an American baseball player, but not as likely that someone would think of him as an African American baseball player? Isn't it more likely that he is thought of as a baseball player in general than any national or ethnic subcategory? Is someone from Los Angeles more likely to be though of as someone from Los Angeles, someone from California, or someone from the United States? Or just someone? This is an impossible and counter-intuitive rule as it asks us to 1.) deliberately create redundant categories, 2.) based on something that seems virtually impossible to know in almost all cases. -Justin (koavf)❤T☮C☺M☯ 07:54, 13 May 2008 (UTC)

You've hit the nail on the head without realising it. The secondary sub-cat rule works in the way that you explained it, but as a way of helping people find things in a category that they would expect not ordering people where to file stuff. Most people would think of him as a baseball player first and then a African-American Baseball player second. The Secondary sub-cat rule is there to address editors who feel the need to categorise things down to the Nth degree. If another editor has filed him away in the black baseball players category but he isn't in the baseball players you can invoke the secondary categorization rule and place him in the baseball category as well. Think of it as a rule that solves disputes and lets both points of view win. Hope that makes sense. - X201 (talk) 14:38, 13 May 2008 (UTC)

No That does not solve it: how do you know that he is thought of as a baseball player first and an African American baseball player second? Where is an American baseball player? Where is American? What about the teams for which he played? Is he "more likely" to be thought of as a Brave or an African American baseball player (both of which are subcategories of American baseball players; itself a subcategory of baseball players)? This is impossible to know and defeats the purpose of a subcategory, since by definition all members of a subcategory are members of the parent category as well. This doesn't solve the problem, it elucidates it. -Justin (koavf)❤T☮C☺M☯ 22:18, 13 May 2008 (UTC)

This issue has been a problem since the time categories were first implemented. The discussion above, about repopulating large categories is directly related to this issue. The current categorization cannot handle this situation well. On the one hand articles could get put only at the very bottom of the hierarchy. This is problematic because people may be just as likely or even more likely to want to browse through a complete list at a higher level. If all the higher levels get populated, then articles fill up with multiple categories from the same hierarchy. The current compromise was worked out to try and deal with the inadequacies of this system. I think of the current compromise as an "all or nothing" decision. If some baseball players can be found in Category:American baseball players, than all American baseball players should be similarly categorized. If there are subcategories for every ethnicity of ballplayer, and consensus is to remove articles from the parent category and only populate the ethnicities, then there would not be a duplication. The situation we don't want is to create subcategories where only some people are removed from the parent categories. I really hope we can eventually move towards a more sane system that fully populates these larger categories, and uses category intersection to find the smaller ones. This is in the works. See the sections above for more details... -- ☑ SamuelWantman 05:42, 14 May 2008 (UTC)

[edit] Removing a hidden category from an article or talk page.

Hello, I need help removing a hidden category. A while ago, someone placed Talk:Crush 40 into Category:All articles that may contain original research, which is a hidden category. I have a feeling it was a couple of months ago, when the article did have this problem. Since then, the article was cleaned up and even passed a GA nomination, which states that articles cannot have original research to pass. When I found the category this morning, I couldn't figure out how to remove it. I'd like some help with this, because it really does not belong in that category anymore. Red Phoenix ^{flame of life...}^{protector of all...} 14:08, 13 May 2008 (UTC)

Done It came from {{or}} in a comment. Ugh. -Justin (koavf)❤T☮C☺M☯ 23:41, 13 May 2008 (UTC)

[edit] Semantic categories with magic words

(moved from Wikipedia:Village pump (proposals).)

I tried searching for this idea in Wikipedia and on the web. In Wikipedia i found nothing and on the web i found a couple of academic papers that propose something similar. My idea, however, appears much easier to understand and to implement.

The current categories system in Wikipedia is already better than anything else i've ever seen in any other encyclopedia or digital catalog. Nevertheless they have a weakness - they are not semantic. They aren't so good at telling the kind of relationship between the article or its subject and the category.

In practice, categories today are used for several kinds of relationships. Let's take the city of my birth as an example:

Relationships of the article's subject to categories - "Moscow ...
- ... is a: | | Federal cities of Russia | Capitals in Europe | Hero Cities of the Soviet Union | River cities
- ... is related to: Golden Ring of Russia

Relationships of the Wikipedia article itself to categories - The Wikipedia article Moscow ...
- ... is the main article in: Moscow (encyclopedic category)
- ... is a: All articles with unsourced statements | Articles with unsourced statements since May 2008
- ... is a part of Wikipedia maintenance project:: Wikipedia external links cleanup

I can think of a few more types of relationships, but no more than ten.

Many pages have been written about making MediaWiki closer to a relational database model. OmegaWiki seems to progress slowly in that direction, but my impression is that people who don't understand database normalization will find it very complicated. Semantic MediaWiki is somewhat similar to my proposal, but it appears to me as very repetitive and overly complicated (and its websites appear to be dead). My proposal is more natural for Wikipedia, which thrives on free-form linking and editing.

Category:Hidden categories solve this problem partially, but they are more about presentation than about meaning (although in practice it us usually used on maintenance categories.) I propose to define semantic categories similarly to hidden categories - with a magic word. The advantages of using a magic word:

It requires only one edit (unlike in Semantic MediaWiki, in which every capital city would have to be linked).
No need to define new keywords: users only need to know [[Category:]], just like today.
Doesn't seem to be harder to implement than hidden categories (i'm not a MediaWiki developer, but it really appears to be very easy).

I also believe that implementation of this idea it will diminish these heated arguments:

about proper sorting of categories ([[Category:Bla| ]] or [[Category:Bla|*]] or {{catmore}}?)
double categorization (should the Welsh language article be in Category:Brythonic languages or only in [[:Category:Welsh language?).

So, any other thoughts? --Amir E. Aharoni (talk) 19:28, 14 May 2008 (UTC)

What are these magic words you propose? What would they do? How will they be used? What will be different? What will be the same? Sorry, I don't understand what you are proposing. -- ☑ SamuelWantman 06:13, 15 May 2008 (UTC)

First, the functional things:

What will be the same - articles would still be included in categories using the current syntax: [[Category:Capitals]]

What will be different - it will be easier for computers (and humans) to understand whether the members of a category have a relationship of article name IS A category name (Moscow is a capital city) or a relationship of article name IS RELATED TO THE TOPIC OF category name (Moscow is related to the Golden Ring of Russia).

Currently in the bottom of every page you have a gray "Categories" field which says:

Categories: Cat1 | Cat2 | Cat3

I propose that it will say:

Moscow belongs to: Capital cities | Host cities of the Summer Olympic Games"

Moscow is related to: Golden Ring of Russia

This Wikipedia article belongs to: All articles with unsourced statements | Articles with unsourced statements since May 2008

(Feel free to think of better wording!)

You can think of wildly positive implications to this simple change. It will be much easier to query Wikipedia as if it was a relational database. Currently you can do automated tasks with categories, but you are limited. There is no way for a bot to tell the semantic difference between Category:Hills of Russia and Category:Battle of Stalingrad. For a human it might be easy to tell that Mamayev Kurgan is a hill in Russia that is related to the Battle of Stalingrad. But there are many hills in Russia, but only one Battle of Stalingrad and as far as a computer program is concerned, both have many articles in them. But if you mark Category:Hills of Russia with the proposed magic word __IS_A__ and Category:Battle of Stalingrad with the proposed magic word __RELATED_TO__, then the computer will be able to tell something about it.

(Again, feel free to propose better names for the magic words!)

This is not a full solution, such as Metaweb's Freebase or OmegaWiki. It's a very simple solution which will be very easy to implement and use. --Amir E. Aharoni (talk) 10:56, 15 May 2008 (UTC)

Looks a great idea in my opinion. Something to take to the developers at Bugzilla perhaps (though my experience is that they're not always too open-minded when it comes to category-related feature requests). In the longer term though, I'd like to see something more substantial undertaken in the semantic direction, along the lines of SemiMedia Wiki and this.--Kotniski (talk) 11:41, 15 May 2008 (UTC)

Doesn't the grammatical distinction usually indicate the difference between a compound name and a group of things? I admit I can think of difficulties, such as where the same form is used for both. DGG (talk) 17:06, 21 May 2008 (UTC)

Grammatical distinction is not completely consistent. Besides, it may be easy for humans to understand it, but not for computers. --Amir E. Aharoni (talk) 17:38, 21 May 2008 (UTC)

[edit] Main article to category relationship guidelines

I would like to seek some guidance as to the criteria by which the main article of any given category should have a relationship to the subcategories the head category covers. I.e. if the Main category A consists of sub-categories B, C, D and E, then the Main article for the category A needs to cover in some substance the sub-categories B, C, D and E at least as sub-sections. Is this assumption correct? (posting in two places as was unsure where this is best posted)--mrg3105 (comms) ♠♥♦♣ 23:25, 20 May 2008 (UTC)

This is an interesting question. I'd be interested is hearing of examples where this is the case, and examples where it isn't. I suspect that the distinctions that help organize articles into a good categories are not always the same distinctions that help organize information into a good article. Often the subcategories are individual instances of the topic, for example, films organized by year, by genre, etc... This makes sense for organizing thousands of articles about films. It may translate into into sections about film genre and film history in the film article, but it may not. -- ☑ SamuelWantman 09:21, 21 May 2008 (UTC)

[edit] Categorisation of images by user

I'd like to request discussion about whether and/or when images should be categorised by user (see Category:User-created images).

I can think of two types of relationships that could be categorised by an "images by user" category:

Images created by a particular user (for example, photographs taken or diagrams developed by the user).
Images uploaded by a particular user, irrespective of whether they created the image (for example, logos uploaded by the user but belonging to someone else).

The latter relationship does not seem to me to be worthy of categorisation; editors who simply wish to keep track of their uploads can maintain a list on a user page or subpage. (The upload log provides a similar functionality.) The former relationship may be a more useful basis for categorisation, but images created by Wikipedia users should generally be uploaded to Wikimedia Commons (assuming they are released under a free license, which most are), and there seems little reason to keep and categorise them on Wikipedia.

Some relevant CFD discussions:

Any thoughts? –Black Falcon ^(Talk) 17:41, 22 May 2008 (UTC)

The long standing consensus has been that categories are an encyclopedic resource created and shared by all users for the benefit of all users. Self-referential categories have existed for maintenance purposes, and were recently made hidden. The categories you describe have never survived CFD as far as I know, nor should they in my opinion. They do not fit the stated purposes of categories, which is to help general readers browse through pages based on subject matter. Many users create lists of the images they upload in their own user-space, and there have been tools that create lists of these images (I'm not sure which tools are currently working). Another big problem with these categories is that once an image is created, anyone can modify it. I often edit images that I find. Wikis are a collaborative effort. I could only imagine the first case surviving CFD if the user were to become a notable photographer, in which case it wouldn't be images created by a particular user, but images created by a notable photographer. I'm rather surprised that some of these categories have existed for as long as they have. -- ☑ SamuelWantman 09:52, 23 May 2008 (UTC)

[edit] New template request

Can anyone give me a hand in creating the categorisation equivalent of this template {{Maintained|{{User|username}}}}?--mrg3105 (comms) ♠♥♦♣ 09:47, 24 May 2008 (UTC)

[edit] Categories for albums by producers

While some would like to see these categories banned, I would like to see them used accurately, and I'm wondering if people here have any thoughts on the matter. I feel that if an album is produced (or co-produced) by a person, we can put that album in the category. But if that album has only one song that was produced by that person, it doesn't belong in the albums produced by category. Am I right? Wrong? Doesn't much matter? If the practice is a bad one, how do we enforce it? Also see Category talk:Albums by producer if you're interested in the subject. -Freekee (talk) 02:44, 28 May 2008 (UTC)

[edit] Category naming

There seems to be a confusion about the purpose for existing of the article titles and categories.

Article titles and categories are not the same

The article title informs the reader about the subject of the article

The category serves to combine many articles on a similar class, attribute, relationship, function, event or other commonality between the articles it contains.

The article title follows naming conventions that deal only with its relationship to the article content expressed in standard English use of nouns, adjectives, verbs, etc.

E.g. Article Little red riding hood

The category name reflects the hierarchy of its position within the categorisation scheme

Category:Female child characters in print fiction

Category:Female characters in print fiction

Category:Characters in print fiction

Category:Printed fiction

Category:Fiction

In effect the job of categorisation entails structuring the category names to reflect their hierarchical position by ideally using a single word to reflect class, attribute, relationship, function, event or other commonality contraction or expansion.

Besides that the categoriser needs to consider the fact that the English language is read left to right. This means the reader has an instinctive ability to disambiguate categories left to right from a specific term (what is being sought) to the general umbrella term (where it is found).

So, replace this

Categories follow the same general naming conventions as articles; for example, common nouns are not capitalized. For specific conventions related to categories, see Wikipedia:Naming conventions (categories). Whatever categories you add, make sure they do not implicitly violate the neutral point of view policy. If the nature of something is in dispute (e.g., if an event is considered a war crime), you may want to avoid labelling it or mark the categorization as disputed. ~~Most naming, however, is straightforward.~~

with this

Category naming does not follow general naming conventions of articles, but its own convention that needs to reflect the hierarchical or network structure into which a category is being created. Ideally, the category name should have as many words (except the, of, in, etc.) as there are levels of parent categories between it, and a Project root category which usually has one word. The category name should contain the most specific term referring to the subjects of the articles on the left (first position), and the most general root category word on the right (last position). Intermediary words are derived from higher order categories, arranged to suit the English reading in the best way possible to reflect category stratification rather then strict application of grammar. Do not implicitly violate the neutral point of view policy. If the nature of something is in dispute, avoid using it, or mark the category as disputed.

Comments?--mrg3105 (comms) ♠♥♦♣ 05:55, 30 May 2008 (UTC)