Personalized Recommendation System Based on Deep Reinforcement Learning

With the rapid development of Internet technology, personalized recommendation systems have become increasingly important in fields such as e-commerce, social media, and online entertainment. However, accurately capturing the dynamically changing interests of users and providing high-quality recommendations still remain the major challenges in the field of recommendation systems. This paper proposes a personalized recommendation system based on deep reinforcement learning, and adopts the Actor-Critic architecture to optimize the recommendation strategy. This paper conducts a detailed analysis of the impacts of different reward normalization strategies, experience replay, and state representation strategies on the performance of the model. Through experiments conducted on the MovieLens ml-1m dataset, we have verified the effectiveness of this method. This method not only outperforms existing methods in terms of total reward, Q-loss, and precision, but also demonstrates better stability and scalability when dealing with large-scale data and complex user behaviour patterns. It also explores how to further improve the accuracy of the recommendation system and user satisfaction, providing new perspectives and methods for future research on recommendation systems.