One example of an unexpectedly rich and deep tagging ontology is the Danbooru "Anime" image board [NSFW] https://danbooru.donmai.us/
They have tag aliases, meta-tags and so-called "tag implications".
The last one is basically sub-tags but with more flexibility and dead simple to implement: if A implicates B, then tagging an image with A will automatically tag it with B. So you can tag "American Male Novelist", and then the system will automatically add "American", "Male", "Novelist", "Writer", etc. (after such implications were added).
It much easier than Wikipedia's categories, but Wikipedia's way is of course intentional because categories is meant to have a stronger hierarchy than mere tags.
But yeah, I've always been impressed by how detailed and specific their tags were. That's A LOT of work! The power of porn, I guess. And as an added bonus, those tags now power AIs able to generate custom hentai on demand!
If you want to find some images by tags, you better stick with more generic ones.
(It is of course based on the tagging system: every post is tagged by its "safeness" level.)
There was some drama about someone training a Stable-Diffusion-alike by ripping their dataset that brought it to my attention.
Anyone who's into Anime (not just for hentai) probably knows.