Students’ Engagement in Learning Programming Using a GenAI-Driven Chatbot: Insights from Sentiment and Epistemic Network Analysis

Abstract

The study examines the sentiment-related engagement of
students interacting with a GenAI-driven chatbot in
learning programming, focusing on two distinct groups:
students with score improvement (Group A) and those without
score improvement (Group B). This study leveraged sentiment
analysis and epistemic network analysis (ENA) to examine
the relationships between sentiment expressions and
learning behaviors. The findings reveal that students in
Group A exhibited greater diversity in sentiment
expressions, characterized by more constructive use of
feedback, whereas students in Group B predominantly
displayed neutral and negative sentiments with limited
feedback utilization, positing lower engagement. These
results provide insights into how students’ sentiment
patterns relate to learning strategies and highlight the
potential of ENA in capturing complex interactions between
affective expressions and cognitive behaviors in learning
programming.