Automatically suggest categories when creating a new article
Problem: It is often difficult to think of what categories a new article should be part of, leading to many categories being incomprehensive and many articles harder to find.
Who would benefit: Editors and readers who value categories.
Proposed solution: Add a function to the process of creating a new article which automatically suggests categories for it to be included in.
More comments: The way the suggested categories would be chosen could be for example by searching the article title in category titles, or by taking a sentence in the article that fits a generic categorizing pattern ("[article title] is/was X"), and then searching for X. Or perhaps there could be some kind of an algorithm that can look for similar articles (something like the one used for 'suggested articles' on mobile), and then showing the categories those are in.
Oppose Getting an algorithm to do this task well enough to be useful might not be easy, so this seems like a high-effort request for only medium benefit. Less significantly, I'm also a little concerned that editors might start relying on this and stop adding categories once the bot runs out of suggestions. {{u|Sdkb}}talk 04:38, 9 December 2020 (UTC)Reply[reply]
Support This will be very problematic in different languages. But some algorithm should compare categories in different language versions (via Wikidata) Another solution should be suggesting via infobox and some parameters, JAn Dudík (talk) 10:53, 9 December 2020 (UTC)Reply[reply]
Support Great for new editors, and hopefully an opt-out function for those who wish to Lugnuts (talk) 12:28, 9 December 2020 (UTC)Reply[reply]
Support another approach could be a better search function on the category search for some frequently used categories rather than having to get the precise title e.g. 'women' could also pull up 'female'? Kaybeesquared (talk) 13:34, 9 December 2020 (UTC)Reply[reply]
Oppose generally opposed to sinking resources into this system that should just be replaced by Wikidata. It's hard to find categories because people have to figure out what we've called things and what currently exists instead of relying on uniform Wikidata properties, which can be combined and don't depend values pre-existing. Instead of trying to guide them to what we call the categories, just have a form with dates, fields, societies, birthplace, etc. without figuring out whether this category has the country first or country last, or whether it uses "in" or "of," etc. — Rhododendritestalk \\ 14:02, 9 December 2020 (UTC)Reply[reply]
Support But there are same better things to do. Ján Kepler (talk) 14:47, 9 December 2020 (UTC)Reply[reply]
Support This is a great idea. I sometimes pick a random category when creating, then have to search for the correct one and replace it in the article after creation. Oaktree b (talk) 16:35, 9 December 2020 (UTC)Reply[reply]
Weak support Potentially useful but I feel this could become an expensive use of resources to make it work satisfactorily. // Lollipoplollipoplollipop :: talk 05:19, 10 December 2020 (UTC)Reply[reply]
Oppose per Rhododendrites. I'm also highly unconvinced that such a tool would successful rather than laughable/annoying without a significant investment of resources on higher-priority tasks. Our categorisation system is high-effort for low-reward, with it not having a particularly large benefit to readers (who scrolls past the prose/images/text/navboxes to read the categories?) — Bilorv (talk) 01:05, 11 December 2020 (UTC)Reply[reply]
Support Very strong support. It can be based in the page name. BoldLuis (talk) 11:22, 11 December 2020 (UTC)Reply[reply]
Oppose per Rhododendrites and Bilorv, particularly Our categorisation system is high-effort for low-reward. ProcrastinatingReader (talk) 11:25, 11 December 2020 (UTC)Reply[reply]
Opposegenerally opposed to sinking resources into this system that should just be replaced by Wikidata This 1000x. Haste the day. czar 17:33, 11 December 2020 (UTC)Reply[reply]
Support Would be very useful even if the suggestions were limited. So long as they are accurate and relevant. --Mathieugp (talk) 18:54, 11 December 2020 (UTC)Reply[reply]
Oppose This would only create a mess, that would have to be later sorted out by manual labor anyway.-Darwinek (talk) 23:54, 12 December 2020 (UTC)Reply[reply]
Oppose for dewiki, we have a complex categorization system with thematic and object categories in which I don't see an automatism getting useful results. -- hgzh 17:02, 13 December 2020 (UTC)Reply[reply]
Support ~ Amory(u • t • c) 19:50, 13 December 2020 (UTC)Reply[reply]
Strong oppose agree with (esp 2nd) point by Sdkb [who needs to fix signature code ;) ]: potential over-reliance on this feature. Anyway, interested parties can add new categories later if obvious/relevant ones were omitted upon initial post. Philiptdotcom (talk) 14:24, 14 December 2020 (UTC)Reply[reply]
Oppose per the concerns brought up by other opposes. I don't believe this is worth the time or energy to develop, especially for the categorization system which is already debatably subjective to any reader or editor (i.e. there's more than one right answer to categorizing pages), and enforcing a standard for categorization at this point in time could be a nightmare. And I know that they're only suggestions, but they would be the only suggestions being offered. Utopes (talk) 19:44, 15 December 2020 (UTC)Reply[reply]
Support It is good to learn that ORES already has such a capability. It could be a good incentive to train it to support more languages when there would be a UI using this underlying thing Base (talk) 20:13, 17 December 2020 (UTC)Reply[reply]
Support Stryn (talk) 18:25, 8 December 2020 (UTC)
Oppose Getting an algorithm to do this task well enough to be useful might not be easy, so this seems like a high-effort request for only medium benefit. Less significantly, I'm also a little concerned that editors might start relying on this and stop adding categories once the bot runs out of suggestions. {{u|Sdkb}} talk 04:38, 9 December 2020 (UTC)
Weak support Potentially useful but I feel this could become an expensive use of resources to make it work satisfactorily. // Lollipoplollipoplollipop :: talk 05:19, 10 December 2020 (UTC)
Strong oppose agree with (esp 2nd) point by Sdkb [who needs to fix signature code ;) ]: potential over-reliance on this feature. Anyway, interested parties can add new categories later if obvious/relevant ones were omitted upon initial post. Philiptdotcom (talk) 14:24, 14 December 2020 (UTC)
