Utilizing knowledge graph and student testing behavior data for personalized exercise recommendation

Personalized exercise recommendation plays an important role in boosting the study performance of students. However, recent studies for personalized exercise recommendation only take the learning status of a student in recommendation and fail to take the prerequisite relationships among knowledge points into account which represent a reasonable learning sequence of these knowledge points during a study procedure. To the best of knowledge, in this paper, we make the first attempt employing both of the learning status of a student and the prerequisite dependencies among knowledge points to enhance the effectiveness in personalized exercise recommendation. A real-case evaluation confirms the effectiveness of our personalized exercise recommendation algorithm in terms of recommendation precision and diversity.