Research and application of tag-based recommendation algorithm based on reinforcement learning
基于标签的强化学习推荐算法研究与应用

In order to solve instability problem in the performance of collaborative filtering recommendation algorithms caused by the data sparseness, this paper proposed an algorithm called tag informed reinforcement learning recommendation model (TIRLR) in framework of reinforcement learning. This paper used the tags to simulate user profiles to construct substantial personalized data, and combined simulated data and historical data to collaborative filtering recommendation. Experimental results show that TIRLR can effectively enhance the performance of collaborative filtering algorithms, and it can get the best result by combining simulated data and historical data.