Our work, funded by Google, explores how we can support personalised learning in MOOCS by automatically classifying and labelling discussion forum topics. This work helps learners navigate large amounts of discussion data, by helping them focus on the topics that are of interest to them, and helps academics by allowing them to see those topics that are attracting the most questions and discussion at a particular point in time.
Recent results from this work, exploring our visualisation approach, has been accepted for the Ninth International Conference on Educational Data Mining (EDM 2016).
T. Atapattu, K. Falkner, and H. Tarmazdi. Topic-wise Classification of MOOC Discussions: A Visual Analytics Approach. Accepted for the Ninth International Conference on Educational Data Mining (EDM 2016).
With a goal of better understanding the online discourse within the Massive Open Online Course (MOOC) context, this paper presents an open source visualisation dashboard developed to identify and classify emergent discussion topics (or themes). As an extension to the authors’ previous work in identifying key topics from MOOC discussion contents, this work visualises lecture-related discussions as a graph of relationships between topics and threads. We demonstrate the visualisation using three popular MOOCs offered during 2013. This work facilitates the course staff to locate and navigate the most influential topic clusters as well as the discussions that require intervention by connecting the topics with the corresponding weekly lectures. Further, we demonstrate how our interactive visualisation can be used to explore correlations between discussion topics and other variables such as views, posts, votes, instructor intervention, and time.