Automatic Evaluation of Concept Maps From an Immersive Virtual World for Ecosystem Science Learning

• In Event: 66.032 - Reenvisioning Assessment and Approaches in Learning Technologies

Abstract

Concept mapping is an effective way of making group thinking visible during collaborative science activities. Scaffolding the construction of causal maps from an immersive virtual world enables students to represent the complex causal relationships in the simulation. This paper details our automatic coding of claims and evidence used by students in their concept maps, with a focus on how maps vary by the sources of evidence they depend on. We analyze structural differences between students’ concept maps, the different types of claims made, and the roles of different types of evidence present in the world. Future work will synthesize this automated coding with log file data to see longitudinally how groups’ thinking changes over time.

Authors

• Joseph M. Reilly, Harvard University

• Shari J. Metcalf, Harvard University

• Jamie Studwell, Harvard University

• Amy M. Kamarainen, Harvard Graduate School of Education

• Tina A. Grotzer, Harvard University

• Christopher J. Dede, Harvard University