Abstract

The COVID-19 pandemic has led to a surge in the number of online educational videos available to learners. However, manually extracting topics from these videos can be time-consuming and inefficient. In this paper, we introduce SpeeKG, a system that automatically translates transcripts from educational videos into concept maps, providing a structural and conceptual knowledge. We evaluated our system using real-world lecture videos collected from a popular video-sharing platform in terms of inter-annotator agreement. Our results demonstrate the potential of SpeeKG to facilitate efficient and accurate topic extraction from educational videos.

Highlights