Clustering Heterogeneous Ink Data
for Classroom Presenter
by
Jim Li and Makara Kov
In this talk, we illustrate clustering techniques used to group heterogeneous ink data submitted by students for the Classroom Presenter. The clustering algorithms are based on stroke information of the slides, which could include texts, diagrams or graphs. These techniques allow the instructor to quickly identify similar submissions and measure students' understanding based on the grouping results. The overall system architecture, implementations, test results and analysis of our research will be described in this talk.
Advised by Richard Anderson
CSE 403
Wednesday
April 26, 2006
3:30 - 4:20 pm