A popular service without/with a bad recommendation system?

Question

Algorithms

Hello,
As one of the options for the thesis, there is a desire to write a recommender system. But for what / what service? Perhaps the Toaster community will help me choose a service (foreign) in which you personally lack a recommendation system or the current system is simply terrible?

Answer 1

You can hardly surpass the recommendation system of any popular service, even the most lagging one. Still, there are no fools working there.
First of all, look where you can parse / pull data from, because companies usually carefully monitor that no one can use their work for free.
About 7-8 years ago, Netflix held an open competition to improve the recommendations of their films, laid out data for analysis. Google it.
There is also this thing: https://www.kaggle.com/competitions

Answer 2

bad / good - it's all relative, take any service, make it better and the previous one will become bad).
Why do you need overseas services?
if you really want to build a recommendation system, then you need to start at least from the data that is provided to you, does the service open data according to users' preferences? to then build some recommendations?
in general, do not bother, take VKshechka, and make recommendations for posts / news, for example.