CoGE: A Data-Based Framework for Assessing Collaborative Group Engagement in Computer-supported Learning Environments

Abstract To assess collaborative group engagement within computer-supported collaborative learning (CSCL) en-vironments, we introduce a comprehensive, data-based approach, the Collaborative Group Engagement (CoGE) framework. It integrates natural language processing and computer vision in a structured, flowchart-based process. The CoGE framework aims to ensure consistent assessments of behavioral, metacognitive, and socio-emotional dimensions of group engagement. Central to the framework is a temporal segmentation strategy, enabling aligned and contextualized analysis across data dimensions. The time segmentation is used in tailored visualizations of multidimensional data, facilitating nuanced interpretation and informed decision-making through raters. The CoGE framework leverages semi-automation of processes to enhance objectivity and reliability, addressing challenges associated with traditional, subjective assessments of collaborative engagement. The presented framework improves the assessment of group engagement, offering a scalable, objective, and multidimensional approach. The presented CoGE framework is implemented in a prototype application to test and illustrate its applicability. The application, a dashboard integrates interactive Visual Analytics to support and streamline the interpretation of group engagement, enabling insights, fostering informed data-based interventions, and improving rater efficiency and objectivity. The implemented prototype demonstrates the practical feasibility of the CoGE framework, showcasing its capability to bridge the gap between semi-automated data processing and human-centric evaluation in collaborative learning settings.