Data-driven and human-controlled intelligent recommender agents for digitalized education

Abstract

In the environment of e-learning many approaches for recommender systems have been proposed. They are based on data or human driven knowledge, which have different advantages and disadvantages. To combine these two approaches, the data of three elearning courses was analysed. The data-based knowledge was combined with human knowledge of the corresponding course lecturers. The combined knowledge was represented in a Bayesian network to generate recommendations for students attendingthese e-learning courses. A prototype was implemented to simulate one of these coursesand to propose recommendations to test persons for an evaluation.