Estimating Student Learning Ability from Massive Open Online Courses

Abstract

Massive open online courses (MOOC) provide a possible way for students to learn knowledge by themselves. Since the number of enrollments for each course is much larger than traditional in-person courses, it is hard for teaching faculty to master the learning ability of each student. However, estimating student learning ability is of great importance for delivering course content. Towards this goal, in this paper, we provide a novel way to estimate personalized learning ability for each student from their answering records in an exam by applying machine learning techniques, called truth discovery, which can automatically estimate a weight for each student and infer answers of questions. The weight can be considered as the student learning ability. The experimental results demonstrate the effectiveness of the utilized truth discovery approach for estimating student learning ability.