A Predictive Model for Video Lectures Classification

In the educational context, it is important to provide stu-dents with learning resources, such as tutorials, video lec-tures, and educational games to help their learning process, especially when they are not in the school and have diffi-culties or doubts. In these situations, recommendation sys-tems may be used to suggest learning resources for students, avoiding, for instance, the task of making the manual pro-cess of searching and selecting resources. Most generic rec-ommendation systems for video lectures use viewing history of the user to make recommendations of videos, which are in according of the user interests. In the educational con-text, other factors must be considered, the video should not only be of interest to the student, as it will be used as a learning resource for the main purpose of helping the stu-dent to learn a particular subject or clarify doubts. Hence, in this paper we evaluated three classiers and propose a predictive model to classify video lectures according to their quality. We applied machine-learning algorithms on a set of video lectures by classied students according to some qual-ity requirements. We conducted an experiment and prelim-inary results indicate good quality of the selected prediction model. 1