For example when you hit the like / upvote button it affects the future recommendations you receive.
At a high level how do sites determine all of this information?
Every time you hit the like/dislike button does it automatically make a call to the backend and update your recommendations list?
Or does it wait a bit before it does that in case you hit like by accident? What would happen if you just spam and keep liking and unliking a particular video or post?
this is done by many ways:
FaceBook:
1- when you stay reading the post for a long time or hit like, then you are interested about the content that is offered in this post.maybe in the database there are categories for the data related to posts and offers according to it. also facebook has strong backend which can analyze photos and according to it's data, it may save it in the DB. this is done by image recognition using python. most people also allow facebook accessabilty for apps on their phones, this allows facebook to collect data about your interests. recommendations also based on your FB data, like age, location, friends and other data. when you admire something, the next time you refesh or opens facebook, a call to backend is made and then it returns data according to some cases according to user data and specifications.
Reddit this mostly depends on text analysis because it is mostly reading posts. and by the same way like facebook it shows data
YouTube here it maybe a little bit different because here the backend will analyze the videos you watch, channels, and other videos related json data (if you have dealed with yt_dl you will understand me)
you asked about when will the user preferences be updated?. i think this is done when you refresh or open the app again