Video segmentation as an example for elaborating design patterns through empirical studies

Abstract

While pattern mining is characterized by empirical examinations the applied solutions propagated in the patterns are only rarely the subject or the result of empirical tests. However, confidence that a pattern will serve its purpose can only be obtained through testing and using them in practice. This article uses the pattern VIDEO SEGMENTATION as a use case to present an iterative approach on how to integrate iterative empirical tests into the elaboration of design patterns. This approach aims to explore the design space and specifies the context of a pattern. With the approach presented here, we want to encourage researchers from different disciplines to communicate their empirical research results in the form of design patterns. Using learning analytics methods, two studies were conducted for this purpose. In the first study 10 popular educational YouTube channels were analysed regarding the length and sequence of 4.136 videos. By using a cluster analysis three groups of videos of different length could be identified: videos under 8 minutes playing time, videos between 8 and 20 minutes duration and longer videos. For the second study 22 participants were split up into two groups using a video player supporting segmented videos in two different ways, and a control group watching a non-segmented video. Segmented videos with sequences of up to 10 minutes resulted in higher learning effects than the non-segmented version of the same video.