Abstract:

Analyzing scientific literature on the impact of emotions in e-learning highlights the importance of assessing students' emotional states and their role during online learning processes. This assessment often relies on physiological sensors, which may not be accessible to all institutions or acceptable to all students. Therefore, detecting students' emotional states in an online learning environment remains a complex challenge requiring a fundamental understanding of the role of emotions in such environments.

This paper explores the potential benefits of using self-reported surveys to identify students' affective states and improve their learning experience by responding to specific emotional states during their learning progress. Our objectives include identifying emotional states through surveys, pinpointing frequently selected emotion-descriptive words, analyzing survey results, and developing a Bayesian Network (BN) model for automatic emotional state analysis. This BN model aims to facilitate real-time interactions within an adaptive learning system by autonomously assessing students' emotions without direct educator intervention.

We present findings from an emotion identification survey conducted among engineering students in an e-learning course provided in an adaptive learning environment. Practically, we plan to implement the BN model within the aforementioned environment to interact with students through real-time pop-up messages, identifying and responding to their emotional states. Its generic character structure allows for changes in the set of analyzed words, ensuring its broad applicability. The challenges of identifying emotions during different online learning activities and the question of the optimal word selection remain for future exploration.

Effective identification of students' emotional states is expected to enhance online learning, improving emotional well-being, motivation, and individual success.

Keywords:

e-learning, Emotion Identification, Automatic Detection Techniques, Bayesian Networks.