Abstract:

Since 2008, the Junior Study program of the University of Rostock offers study orientation and preparation during the school time for pupils of German high schools. For this, regular lectures are recorded and then provided via an online learning platform. With 250 young students in winter semester 2017/2018, Junior Study program of the University of Rostock is one of the biggest and most successful early studies offerings in Germany.

An empirical study, which we conducted at the end of the last semester, revealed amongst other things attention and motivation problems of students while watching 90-minute university lecture videos. The same drawbacks are observed in the plenty of massive open online courses that are mainly based on recorded video lectures. Using webcams or cameras of mobile devices for capturing, free software for editing, and online platforms like YouTube for free sharing, instructional videos can be easily and favourably created and distributed. Hence, there’s a huge selection of videos relating any topic and it’s growing every day. Although videos became extremely popular as a source for formal and informal learning processes, especially long videos are ill-suited for the required lifelong learning just in time and on demand. The learning success-oriented selection of appropriate video segments and their arrangement towards individual learning targets are inhibited. While watching a video, already known facts are unnecessarily repeated or rather can only be skipped manually. The same problems occur when attempting to selectively rewatch certain video sections.

While providers of massive open online courses therefore tend to produce specifically designed videos, which is expensive and time-consuming, we follow another approach which includes the semantic processing of existing video materials. For this purpose, we developed a web app called “PyQuizmaker” that enables users to easily embed quizzes into online videos. Based on the Behaviorism theories of Programmed Instruction by B. F. Skinner and Branched Programs by Norman Crowder, follow-up activities can be specified, that are triggered in dependence of the learner’s answers. Based on an automatic evaluation of the user’s current knowledge, the video will adapt according specified behaviours. For example, if questions were answered incorrectly, the video may jump back to the sections which explain corresponding issues. Otherwise, prepended quizzes may check if the knowledge to be conveyed is already available. In this case, the video will automatically skip related sections. Regularly integrated multiple choice tests can be used as learning control and attention grabber. At the same time, a direct positive feedback to given correct answers will motivate the student and reinforce the learning effect. Moreover, authors can add some additional learning materials in terms of text messages, attached files, or external websites in order to show alternative learning paths, to deepen or complement knowledge.

In combination with the feature of explicitly defining video sections with attached tag words, all these information serves as a basis for videos becoming machine-readable. This establishes a wealth of application possibilities like scanning videos for specific topics, automatic linking to related content, or fully automatic, user-centered compilation of individual learning progress-oriented video sections.

Keywords:

Instructional videos, adaptive learning, Behaviorism, self-assessment.