Top-N Knowledge Concept Recommendations in MOOCs Using a Neural Co-Attention Model

Abstract

Massive Open Online Courses (MOOCs), which provide learners with a large-scale, open-access learning opportunity, have drawn a lot of attention recently. The amount of information available on MOOCs is increasing, making it challenging for learners to choose the best course materials and leading to low learning efficiency and high dropout rates. To address these problems, Recommendation Systems (RSs) have been researched as a direct approach of delivering educational content to learners while also attracting their interest. However, a course often consists of various learning concepts, each of which covers a different topic. Explicitly proposing courses can cause the lack of learners' attention to a particular knowledge concept. We introduce a Top-N Knowledge Concept Recommendations in MOOCs using a Neural Co-Attention Model, called NCO-A, that integrates significant heterogeneous data with recommendations based on knowledge concepts. The NCO-A model's effectiveness has been proven by extensive experiments on three real-world datasets.