Generating Automated Feedback and Hints for Algebra Problems Based on Directed Networks

• In Event: Exploring Synergies: Teacher-AI Collaboration in K-12 STEM

Abstract

A major challenge in generating automated feedback in education is the lack of learner models that capture student knowledge and problem-solving processes. This study built a directed network-based student model to generate automated feedback for the Algebra domain using the Bridge to Algebra 2006-2007 data set containing 3,679,140 problem-solving steps taken by 1,135 learners. Three types of automated feedback were generated with a performance rate of 89% on the test set. The study advances our understanding of student modeling for Algebra problems using network analysis. It shows the potential of directed networks to analyze process data and generate adaptive feedback, applicable to other well-defined domains.

Authors

• Yajie Song, McGill University

• Maria Cutumisu, McGill University