Optimisation of UCB algorithm based on cultural content orientation of film and television in the digital era

Abstract

To improve the effect of the upper confidence bound (UCB) algorithm in the recommendation of online courses of film and television culture, the paper proposes the recommendation method with time-varying Linucb. Firstly, the time-varying Linucb is introduced, and the UCB is optimised by using the attention mechanism and the short-term and short-term memory network. The results show that the recommendation accuracy of the improved model reaches up to 93%, and the novelty is basically stable at 70%. Compared with UCB, the average course viewing time of users has been extended by two hours, and the average course registration rate has remained stable at over 84%. This indicates that the improved recommendation model has excavated the diverse learning needs of users and can provide accurate course recommendation services for users, which is conducive to optimising the effectiveness of film and television cultural education.