Snippet Generation for MOOC discussion forums

Abstract

Massive Open Online Course (MOOC) platforms such as Coursera, edX and Udacity have begun to integrate education technology and learning analytics applications to let instructors effectively manage their class and collect feedback on their class performance. A key aspect of MOOC class management involves two-way instructor-student interaction via discussion forum to catalyse learning. However, due to large class sizes of typical MOOCs, instructors are unable to participate in all student discussions or answer all their queries. In this thesis, we address this problem by proposing an end-to-end solution for managing large-scale discussion forums. Our solution is a web-based instructor dashboard which uses a content based recommender system to rank the threads in an order that helps the instructors. We architect a pipeline where a web scraper collates, in real-time, discussion threads posted on the forum and feeds pre-trained machine learning models. The dashboard then renders the re-ranked threads.

 

Members