Design a Reinforcement Learning Recommender (Bandit approach)

A recommender that treats each user impression as a bandit trial, picks an item to show, observes immediate reward (click or not), and...