Hybrid Graph Neural Networks with LSTM Attention Mechanism for Recommendation Systems in MOOCs

Massive Open Online Courses (MOOC) represent a relatively recent development in the educational landscape, rapidly gaining popularity and drawing research attention. Transforming the traditional approach to education, MOOCs provide learners with flexible and accessible avenues for acquiring knowledge. However, the sheer abundance of courses available can overwhelm users. Recommending relevant courses to learners remains a complicated challenge, impacting engagement and completion rates. Conventional recommendation systems often struggle to capture MOOCs' dynamic and interconnected nature. This paper examines the application of Hybrid Graph Neural Networks with a Long Short-Term Memory attention mechanism (HGNN-LSTM) to enhance recommendation systems for MOOCs. By leveraging learner behavior data, these approaches examine and predict learning activities, discern temporal relationships among courses through Adam optimization algorithms, and ultimately enhance the accuracy of recommendations. We illustrate that HGNN-LSTM adeptly captures hidden linkages between courses, resulting in improved automatic course classification and a reduction in the burden of course maintenance. The paper concludes with an analysis of challenges, identified gaps, and suggestions for potential future research directions.