An educational video scoring scheme for level and topic-based recommendation

Recently, a huge amount of educational video contents have become available due to diverse IT technologies and rapidly growing demand for online learning. Popular resources for various high quality educational contents include OCW (OpenCourseWare), MOOC (Massive Open Online Course) and TED (Technology, Entertainment, Design). Moreover, learners can easily access such contents via the Internet by using common video sharing platforms such as YouTube and Vimeo. However, one critical problem that many users experience in this environment is that it is difficult and time consuming to find out educational contents that is appropriate to them in terms of topic and difficulty level. To solve this problem, in this paper, we propose an educational contents scoring scheme for topic and difficulty level based contents recommendation. The scoring is based on the linguistic difficulty of the video and the topic similarity. To measure the linguistic difficulty, we apply various readability metrics to the video script. To measure the topic similarity between videos, we extract keywords from each video script and then calculate their cosine similarity. We show the effectiveness of our proposed scheme by various experiments.