Detecting Neurodiverse Learners’ Frustration During Educational Gameplay Using Sequential Pattern Mining

• In Event: Trends and Challenges of Artificial Intelligence in Education

Abstract

Detecting and monitoring learners’ frustration is crucial to creating an enjoyable and adaptive learning experience. In this study, we explored the potential of utilizing sequential pattern mining of gameplay logs to detect learners’ frustration. A total of 13 neurodivergent learners were recruited to participate in educational puzzle gameplay, during which we collected self-reported affective experience and game log data. We conducted correlation analysis on self-report affect scores to model the relationship between valence, arousal, and mastery. Subsequently, we employed sequential pattern mining to uncover the sequential behaviors highly correlated with sustained frustration. The findings of the study provide preliminary evidence for the feasibility of using log data to predict frustration in underrepresented populations and create an adaptive game-based learning environment.

Authors

• Nuodi Zhang, Florida State University

• Fengfeng Ke, University of Maryland

• Alex Barrett, Florida State University

• Zlatko Sokolikj, Florida State University