User talk:Multichill/Next generation categories

Examples

Examples in the last statement seem incorrect:

• Category:Amsterdam is not a subcategory of Category:North Holland. Rather, it is one of Category:Municipalities of North Holland - this sets relationship precisely: one of many municipalities in a country.
• Category:Churches is not a subcategory of Category:Buildings. Instead, there's a long chain of Category:Buildings -> Category:Buildings by function -> Category:Religious buildings -> Category:Religious buildings by religion -> Category:Christian buildings -> Category:Churches. Too long to be practical (and seldom reproduced at lower levels), but the relationship is set.

Aren't there better examples? NVO (talk) 02:49, 14 April 2011 (UTC)

I picked this example on purpose. These are subcategories, but there's a very long chain from the parent to the child. One of the symptoms of the current category system. Multichill (talk) 08:22, 14 April 2011 (UTC)

Tags

Isn't it far easier to use tags like "2009", "photograph", "Romania", "horse" for all media and then generate automatically, on the fly, all kinds of categories like "Category:2009 in Romania", "Category:2009 photographs", "Category:Horses of Romania" etc. You could even generate the same categories in multiple languages starting from those basic tags. --Codrin.B (talk) 13:19, 10 October 2012 (UTC)

Complication

One of problems is that we categorize not simple atomic concepts but real photos (audio and video files) which can contain more relevant elements. E. g. the photo can depict together a "blue bus" and a "Catholic church". When we use the atomic categorization, it should produce a "blue church" and "Catholic bus". Some images can have two (or more) locations (viewpoint / observed object; location of the depicted artwork / a place depicted on the artwork). --ŠJů (talk) 22:13, 11 March 2013 (UTC)

some tags can be combined without much error, now I believe the important point is not to generate categories on the fly based on keywords, but rather search by keywords, and indicate which keywords can be used to enhance the search. Eg: you search a sets of media containing 'blue', you'll have the keyword associated in the medias counted, and you'll be able to sort the cross over, and go further, reducting the number of image to select / display. Note: this mainly applies to tag and not 'next gen categories'... Esby (talk) 14:47, 13 March 2013 (UTC)

See "beyond categories"

Some of us have been drafting a page on meta on this topic, meta:Beyond categories. Perhaps you could contribute, if you are interested. Klortho (talk) 19:42, 7 May 2013 (UTC)

This isn't about sth better than categories but the UI they are used for

• Multiple languages could be implemented via (semi-)automatic creation of a) translated redirects and b) (machine-) translated category titles and descriptions that are displayed depending on how the user landed on the page (e.g. from Spanish-language search engine results)
• Efficient intersections/searching can be implemented via a more modern and reasonable UI that allows the user to easily combine categories/concepts or view all images in the subcategories of a category as one wall of images etc. This is not about categories but about how they are used and displayed.
• Structured data could indeed be useful for making the relation between categories more explicit. But it's not really that needed or useful. It could now be done using structured data but many times the title of the subcategory already specifies how it relates to the parent category...and if it doesn't usually a broader subcategory can be created as intermediary. In your example something like "Types of buildings"/"Buildings by type" for example.

Maybe you could edit the page to add something about this. Prototyperspective (talk) 15:30, 12 June 2024 (UTC)