With the rapid development of information technology, the emergence of massive reading data provides users with rich choices, but also brings the problem of information overload. In order to improve the reading experience of users and the accuracy of the recommendation system, the study classifies the reading data based on the artificial bee colony clustering algorithm, and optimizes the clustering effect of user behavior data. At the same time, reinforcement learning is used to construct a personalized recommendation model, based on the Markov Decision Process, and the dynamic optimization of reading recommendation strategy is realized through the state-action-reward mechanism, and finally a new personalized reading recommendation method is proposed. Experimental results show that the recommendation accuracy of this new method reaches 94.37%, and user satisfaction is improved to 91.85%, which is 11.73% higher than the traditional collaborative filtering recommendation method. In addition, the recommendation delay is significantly reduced and the average processing time is shortened by 21.4%. It can be seen that this new personalized reading recommendation method can achieve efficient and accurate personalized service in complex and diverse reading scenarios, which provides strong support for further optimization of the intelligent recommendation system.