Article sort, according to each user's habits to recommend different articles, how to consider performance?

rule (the result of formula calculation is the ranking of topic heat):
clipboard.png

:
clipboard.png

I would like to ask how such a requirement can reduce the consumption of server performance, can you tell me about the solution?


how about not being real-time? Do script calculations at a fixed time every day or several times a day, and put the results in redis. So go directly to redis to get it.


1. The key point of the problem lies not in the user's habit, but in the classification of the article. It is necessary to distinguish
2 from the content, establish the corresponding cache rules according to the type and popularity of the article
3, select and match the users with this type of article, and push

randomly according to the subdivision rules.
Menu