Reinforcement Learning-Based Recommendation with User Reviews on Knowledge Graphs

Introducing knowledge graphs (KGs) into recommendation systems can improve their performance, while reinforcement learning (RL) methods can help utilize graph data for recommendation. We investigate existing RL-based methods for recommendation on KGs, and find that such approaches do not make full use of information from user reviews. Introducing user reviews into a recommendation system can reveal user preferences more deeply and equip a RL agent with a stronger ability to distinguish users' preferences for an item or not, which in turn improves the accuracy of recommendation results. We propose Reinforced Knowledge Graph Reasoning with User Reviews (RKGR-UR) by introducing user reviews into a RL-based recommendation model, which combines a rating prediction task to transform predicted ratings into rewards feedback for the RL agent. Experiments on three real datasets demonstrate the effectiveness of our method.