Topic Segmentation for MOOCs
Abstract
While MOOCs have been on the rise, one main issue with MOOCs is the lack of specific resources to help students succeed in these settings. In particular, while there is a general structure to the courses, there is very little segmentation of sub-topics inside each lecture video. In this paper, we build upon state-of-the-art models to automatically segment each lecture video transcript into sub-topics, allowing content creators to receive feedback on how long they are covering each topic on and also for students as a resource to better follow the material and organize their notes. The segmented topics also makes it more efficient for downstream tasks such as information retrieval on MOOCs data.