Learning Sentiments: A Multifaceted Examination of Chatbot Interactions in Massive Open Online Courses

• In Event: Learner Engagement in Online Environments

Abstract

This study investigates students’ sentiments toward a chatbot in massive open online courses, a largely unexplored area. We employ a comprehensive, multifaceted approach, drawing on various data sources: 1,502 chatbot logs providing direct and transactional insight; 174 student open-ended reflection data offering deeper and qualitative perspectives; and 237 student questionnaire responses forming the basis of a predictive machine learning model. Despite most chatbot log interactions being categorized as neutral, student reflections showed substantial positive and negative feelings. The machine learning model identified crucial factors influencing sentiment, including students’ willingness to use the chatbot, social presence, self-regulation skills, ease of use, and cognitive presence. These findings highlight the value of an integrated examination capturing both immediate interactions and overarching reflective sentiments.

Authors

• Songhee Han, Florida State University

• Hyangeun Ji, University of North Texas

• Jiyoon Jung, Valdosta State University

• Unggi Lee, Korea University

• Min Liu, University of Texas at Austin