Using Clustering to Assist Understanding of Digital Ink in Low Attention Environments

Understanding large amounts of multimedia information is a challenge in many domains – as technology makes its way into the classroom, the classroom becomes one such domain. An instructor has little attention to focus on the technology and is instead focused on teaching. We examine several techniques for clustering digital ink diagrams drawn by students during in-class activities. These diagrams are submitted electronically to the instructor in real-time. The goal of clustering is to allow the instructor to gain an overview of the responses submitted, to quickly assess the level of understanding of the students, and to select “interesting” responses to display and discuss further. We have found that an instructor has difficult achieving these goals even in small classes of size 15 to 20. As the class size increases this task becomes impossible. We find that for some exercises our clustering works surprisingly well and should help to reduce the cognitive load on an instructor.