Multimodal Analysis of Video Collections: Visual Exploration of Presentation Techniques in TED Talks

摘要

While much research in the educational field has revealed many presentation techniques, they often overlap and are even occasionally contradictory. Exploring presentation techniques used in TED Talks could provide evidence for a practical guideline. This study aims to explore the verbal and non-verbal presentation techniques from a collection of TED Talks. However, such analysis is challenging due to the difficulties of analyzing multimodal video collections consisted of frame images, text, and metadata. This paper proposes a visual analytic system to analyze multimodal content in video collections. The system features three views at different levels: the Projection view with novel glyphs to facilitate cluster analysis regarding presentation styles; the Comparison View to present temporal distribution and concurrences of presentation techniques and support intra-cluster analysis; and the Video View to enable contextualized exploration of a video. We conduct a case study with language education experts and university students to provide anecdotal evidence about the effectiveness of our approach, and report new findings about presentation techniques in TED Talks. Quantitative feedback from a user study confirms the usefulness of our visual system for multimodal analysis of video collections.