Research Article
A Collaborative Graph Convolutional Networks and Learning Styles Model for Courses Recommendation
@INPROCEEDINGS{10.1007/978-3-031-24383-7_20, author={Junyi Zhu and Liping Wang and Yanxiu Liu and Ping-Kuo Chen and Guodao Zhang}, title={A Collaborative Graph Convolutional Networks and Learning Styles Model for Courses Recommendation}, proceedings={Collaborative Computing: Networking, Applications and Worksharing. 18th EAI International Conference, CollaborateCom 2022, Hangzhou, China, October 15-16, 2022, Proceedings, Part I}, proceedings_a={COLLABORATECOM}, year={2023}, month={1}, keywords={Graph neural networks Learning styles Course recommendation Collaborative models}, doi={10.1007/978-3-031-24383-7_20} }- Junyi Zhu
Liping Wang
Yanxiu Liu
Ping-Kuo Chen
Guodao Zhang
Year: 2023
A Collaborative Graph Convolutional Networks and Learning Styles Model for Courses Recommendation
COLLABORATECOM
Springer
DOI: 10.1007/978-3-031-24383-7_20
Abstract
With the rise of Massive Open Online Courses (MOOCs) and the deepening of lifelong learning, there is a growing demand for learners to learn on online learning platforms. The vast amount of course resources provides learners with massive and easy access while posing challenges in terms of personalized and precise selection. Traditional recommendation models have room for improvement in performance and interpretability in massive open online course scenarios while under-utilizing the potential interaction signals in user-course interactions and ignoring the impact of the user’s learning style as a learner. In order to solve the above problems, this paper proposes a collaborative graph convolutional networks and learning styles model for courses recommendation (CGCNLS). First, the course prediction rating is obtained by propagating the learner-course interaction information recursively through the graph convolutional networks; further, a course and learning styles matching scale is created to calculate the course learning styles similarity score; finally, the course prediction rating is combined with the course learning styles similarity score to make personalized course recommendations. The experimental results show that the model proposed in this paper can effectively recommend courses for learners and outperforms the baseline approach in terms of Precision, Recall, and NDCG performance metrics.
