Learning Path Recommendation Based on Knowledge Tracing Model and Reinforcement Learning

In recent years, studies on personalized learning path recommendation have drawn much attentions in E-learning area. Most of the existing methods generate the learning path based on learning costs that are formulated manually by education experts. However, this kind of learning costs cannot record the knowledge level change during the learning process and therefore does not accurately reflect the learning situation of the learner. To tackle this problem, we propose a knowledge tracing method which models learners' knowledge level over time, so that the learners' learning situation can be accurately predicted. Then, we propose a learning path recommendation algorithm based on the knowledge tracing model and Reinforcement Learning. A series of experiments have been carried out against learning resource datasets. Experiments results demonstrate that our proposed method can make sound recommendations on appropriate learning paths in terms of accuracy and efficiency.