Longitudinal Data Mining on Aspect-Based Sentiment with Behavioral Context in Game-based Coding of Autistic Learners

• In Event: AERA Roundtable Session Wednesday 11:45 am - Gold 3
  In Roundtable Session: Immersive and Data-Rich STEM Learning Analytics

Abstract

Game-based design activities in virtual reality (VR) have shown promise in helping learners with autism develop computational thinking (CT) competencies. However, limited research examines how affective states—such as frustration or satisfaction—interact with CT behaviors in real-time during VR-based learning. This study addresses this research gap through longitudinal data mining with more than 3,000 time-stamped data points collected from two autistic adolescents designing and coding non-player characters (NPCs). By capturing both sentiment and behaviors, this analysis reveals detailed patterns of emotional alignment with specific CT behaviors over time, informing future VR-based instructional design.

Authors

• Xiaoxue Zhou, University of Maryland

• Fengfeng Ke, University of Maryland

• Nuodi Zhang, Florida State University

• Alex Barrett, Florida State University