Topic-based playlist to improve video lecture accessibility

Learning is considered one of the most important activities for the human being, and many educational institutes are trying to improve the engagement with students through the availability of video lectures. However, the access to video material is not easy as one may think. For millions of people with some form of disabilities the simple act of browsing a video lecture archive represents an insuperable burden. Motivated by the need to improve accessibility to video lecture materials, in this paper we present VLP, which stands for Video Lecture Playlist. The idea is to use low-level audio/video features, video segmentation and OCR analysis to "understand" the content of the video lectures. In this way, students can search for specific topic through keywords and the system browses the entire video lecture archive to find all the pieces of video lectures that cover the searched topic. These pieces are then provided through a playlist, so that a single search and a single playout return a complete view of how the searched topic is covered within the entire archive. A developed prototype shows the feasibility of the approach and results obtained from a survey highlight that students would like to browse video lecture archives with keywords and playlists.