A semi-automatic metadata extraction model and method for video-based e-learning contents

• Published in: Education and information technologies 2019-11, Vol.24 (6), p.3243-3268
• Main Authors: Pal, Saurabh, Pramanik, Pijush Kanti Dutta, Majumdar, Tripti, Choudhury, Prasenjit
• Format: Article
• Language: English
• Subjects: Analysis; Automation; Computational linguistics; Computer Appl. in Social and Behavioral Sciences; Computer Science; Computers and Education; Education; Educational Technology; Electronic Learning; Equipment and supplies; Information Systems Applications (incl.Internet); Language processing; Learning; Metadata; Methods; Natural language interfaces; Online education; Teaching; Test Results; User Interfaces and Human Computer Interaction; Video; Video Technology; Web sites
• ISSN: 1360-2357; 1573-7608
• DOI: 10.1007/s10639-019-09926-y

Video-based learning offers a learner a self-paced, lucid, memorizable, and a flexible way of learning. The availability of abundant educational video materials on the web has certainly abetted an individual’s learning means. But the lack of necessary information about the videos makes it difficult for the learner to search and select the exact video as per his/her requirement and suitability in terms of the learner’s learning capability and the material’s relevancy, difficulty level, etc. Educational video recommendation systems also suffer from a similar problem. Extracting the required metadata, by different means, from the learning videos is a plausible solution. Despite the credible research efforts on video metadata extraction, the problem of educational video metadata extraction has been overlooked. This paper proposes a comprehensive approach to extract educational metadata from a learning video. A semiautomatic mechanism that includes manual and computational approaches is introduced for metadata extraction and to evaluate the values of these metadata. Along with identifying a set of specific metadata attributes from IEEE LOM, few additional attributes are suggested which are imperative to assess the suitability of a video-based learning object in terms of the personalized preference and suitability of a learner. The test results are validated by comparing with the manually extracted metadata by experts, on the same videos. The outcome establishes the promising effectiveness of the approach.