Automatic point of interest detection for open online educational video lectures

摘要

The rise of massive open online courses had as a result an increase in the number of open online educational video lectures on the web, as well as in the number of users who watch them. The present work aims to optimize the searching time within an educational video lecture based on the users' opinion. The research presents a novel procedure for the automatic point of interest detection in an educational video lecture. State of the art algorithms are used to extract terminology from the users' comments and video lectures topics from relevant video transcripts. The topics of each video lecture are assessed based on the terminology resulting from the users' relevant comments. The topic with the best score (adding the keyness value of common topic-related words and terminology words) is selected as the most relevant to the video lecture. Then videos' timestamps that include the selected topic's words are presented to users as the points of interest of the video lecture. Finally, a user evaluation experiment is carried out, the results of which strengthen the reliability of the proposed procedure.