In recent years, with the rapid development of the Internet, the e-commerce platform has become an indispensable part of people’s daily life, which can generate personalized recommendations for users to meet their daily life purchase needs. The research adopts the MCTS method combining deep learning and Monte Carlo tree search algorithm, fuses the strategy network, value network and risk network with the strategy value risk network, and combines the improved MCTS method to construct a self-learning strategy value risk network applied to the ecommerce platform recommendation system. The improved MCTS method is verified through experiments, and the improved MCTS method has high comprehensive performance, compares the recommendation effect of the selflearning strategy value risk network algorithm with the UCT algorithm, and utilizes the hybrid reward function to achieve the balance between the accuracy and diversity of recommended products. The improved MCTS method is tested in terms of coverage, diversity and user satisfaction, and the performance is good, with the highest coverage of 0.91 and the highest diversity of 0.9941. In this paper, the system realizes the diversified recommendation of products to meet the user’s need for recommender systems.
