Leveraging Machine Learning to Automatically Evaluate Cognitive Engagement in Asynchronous Online Discussions

• In Event: LED- Online Interaction

Short Description

This study focuses on developing machine learning models to automatically evaluate cognitive engagement in asynchronous online discussions. To this end, the bidirectional encoder representations from transformers (BERT) was finetuned and trained, resulting in an accuracy of 72%. The developed model was utilized to evaluate a previously uncoded dataset, which was then analyzed in terms of learning clusters and trajectories. This research demonstrates the potential of using BERT for cognitive engagement assessment.

Authors

• Contributor: Jinho Kim, Georgia State University

• Contributor: Yoojin Bae, Georgia State University

• Contributor: Golnoush Haddadian, Georgia State University

• Contributor: Min Kyu Kim, Georgia State University