Personalized Hybrid Recommendation Algorithm for MOOCs Based on Learners’ Dynamic Preferences and Multidimensional Capabilities

In the MOOCs context, learners experience information overload. Thus, it is necessary to improve personalized recommendation algorithms for learners. The current algorithm focuses mainly on learners’ course ratings. However, choice of courses not only based interests and preferences. It also affected by knowledge domains learning capabilities, all which change dynamically over time. Therefore, this study proposes a hybrid combining clustering with collaborative filtering. First, data rating preferences, attribute multidimensional capabilities that match traits are used item response theory. Second, considering preferences time, Ebbinghaus forgetting curve introduced integrating memory weights accuracy interpretation proposed MOOCs. Finally, performance investigated using from Coursera, an internationally renowned platform. experimental results show superior baseline algorithms. Accordingly, relevant suggestions development