Knowledge discovery for course choice decision in Massive Open Online Courses using machine learning approaches. (1st August 2022)

• Record Type: Journal Article
• Authors: Nilashi, Mehrbakhsh; Minaei-Bidgoli, Behrouz; Alghamdi, Abdullah; Alrizq, Mesfer; Alghamdi, Omar; Khan Nayer, Fatima; Aljehane, Nojood O; Khosravi, Arash; Mohd, Saidatulakmal
• Is Part Of: Expert systems with applications. Volume 199(2022)
• Publication Date: 2022-08-01
• Subjects: Educational data mining -- Multi-criteria collaborative filtering -- Knowledge discovery -- Text mining -- Course choice decision -- User' preferences; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2022.117092
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: A multi-criteria collaborative recommender system is proposed for MOOCs. Text mining and fuzzy logic approaches are used for method development. Data collection is performed in Udemy.com . Numerical And textual data are analyzed for method evaluation. The method is effective for MOOCs websites for course recommendations. Abstract: Massive Open Online Courses (MOOCs) provide learners with high-quality and flexible online courses with no limitations regarding time and location. Detecting users' behavior in MOOCs is an important task for course recommendations. Collaborative Filtering (CF) is considered the widely approach in recommender systems to provide a online learner courses according to similar learners' preferences in an e-learning environment. The current research provides a novel framework through machine learning techniques to propose course recommendations in MOOCs according to the uses' preferences and behavior. The method is developed using multi-criteria ratings extracted from users' online reviews. We use Latent Dirichlet Allocation (LDA) for text mining, Decision Trees for decision rule generation, Self-Organizing Map (SOM) for users' reviews on courses and the fuzzy rule-based system for users' preferences prediction. We also adopt a feature selection method to select the most important criteria for users' preferences prediction. The method is evaluated using the data collected from an online learning platform, Udemy. The results showed that the method is able to accurately provide relevant courses to the users tailored to their preferences. The method has the potential to be implemented as a recommendation agent in the MOOC websites for course recommendations.