IMPROVING RECOMMENDER SYSTEMS FOR LEARNING: A DEEP DIVE INTO DESIGNING AND EVALUATING EDUCATIONAL RECOMMENDER SYSTEMS

Abstract

Learning management systems (LMSs) are popular tools that are used in e-learning, however, these systems are still suffering from the lack of personalization. This thesis focuses on designing, developing and evaluating educational recommender systems as one of the tools that can be utilized to enhance the functionality of LMSs with personalization. In this thesis, first, the evaluation of two previously built recommender systems (RUBARS and PLORS) in the areas of learner-centered education and learning object recommendation was conducted. The outcomes of the evaluations showed very promising results and indicated that these systems potentially fill a gap in their respected areas. Next, as the main focus of this thesis, a new recommender system called WEBLORS was designed, developed and evaluated. WEBLORS is an adaptive web based recommender system that aims at providing learners with additional recommended, personalized and relevant learning objects from the web. The evaluation of WEBLORS showed very encouraging results. Based on the results of the evaluation, WEBLORS has a very high potential to help learners by recommending extra personalized recommendations from the web and helping them with information overload by only recommending learning objects relevant to the topic that is being studied and which fits students’ profiles.