An Approach to Automatic Evaluation of Instructional Videos

Instructional video quality evaluation is a heavy task due to the rapid growth of available instructional videos and limited experts. In order to solve this problem, this paper presents an approach to automatically evaluate instructional videos. Instructional video evaluation is different from general video evaluation in that the important characteristics of an instructional video is supporting the understanding of course content. Therefore, a metric of instructional video assessment is established. It includes five indicators, i.e. screen effect, subtitle, duration, format and quality. Image recognition and image processing technologies are used to score the indicators. Automatic evaluation model is constructed by machine learning. Our experimental results show that the automatic evaluation is basically consistent with the results of manual evaluation. Therefore, our approach significantly reduces the burden of instructional video evaluation and helps to improve video-based learning.