Investigating students' dynamic learning emotions and interactions in online discussion

Affected by COVID-19, college students had to study at home in China. Therefore, a great significance is to investigate learners' emotions and interactions in online discussion. Many researchers have studied the emotions and interactions among students. However, most of them have overlooked how students' emotions and interactions vary over time, which is a dynamic process in online asynchronous discussion. Using emotion analysis and temporal network analysis, this paper investigates learners' positive, negative and confused states during learning, and uses a methodological approach to build, visualize, and quantitatively analyze temporal network in each week of the course. Results revealed that the ratio of students' positive states was extremely large at the beginning of the semester, and then decreases while the confused or negative ratios increased. The interactions in each week varied with time and could be divided into three patterns according the characteristics of temporal networks. Finally, the relationships among emotions, interactions, and academic performances were analyzed based on temporal networks, and the data revealed that middle-achieving learners were the major contributors while learning emotional states have a slight effect on learning outcomes. This study might assist teachers to provide timely and effective assistance during early warning, so to achieve the ultimate goal of improving the effect of online learning.