Submission Summary
Share...

Direct link:

Activity Theory Informed Automatic Student Assessment Model in Collaborative Math Learning

Thu, Nov 5, 2:15 to 3:15pm, Hyatt Regency, Floor: 3rd, Studio 1

Short Description

Enabling the formative assessment of students while limiting demands on teachers’ time is a significant concern for technology mediated learning in small groups. In this article, we employ activity theory to categorize traces of student behavior captured from a Virtual Math Teams (VMT) geometry class in an online, synchronous environment. From this, six semantically-grounded measures are generated for each student. Using these, a specific spectral clustering algorithm is designed and coded to identify students who have similar behavior patterns, providing teachers with accurate and timely information about student performance. The accuracy of the spectral clustering approach is evaluated through a comparison with other algorithms, with spectral clustering outperforming other algorithms in assessment accuracy. Further, the combined theoretical and computational approach taken allows for an automated and meaningfully-grounded assessment of student performance, enabling teachers to efficiently offer timely, concrete, and personalized help. In the end, a web-based tool is development that implements this methodology to facilitate teachers’ usage.

Author