SENSEI: A Novel Architecture of Deep Multi-Task Neural Network for Analyzing E-learning Lecturers’ and Students’ Satisfaction

E-learningtechnology is widely acknowledged as a crucial component of education. Researchers conduct extensive studies to analyze lecturers’ and students’ effectiveness and satisfaction of e-learning systems via interaction called usage and usability (i.e., SUS) metrics. In a recent empirical study, users’ interactions with an e-learning environment were examined based on a predefined task model that describes low-level interactions. However, the usage-based and usability metrics do not consider opinions of users related to e-learning. Additionally, machine learning algorithms have been widely used for this purpose to analyze satisfaction via opinions. However, they ignore analyzing users’ satisfaction through usage and SUS metrics, and emotions. In this study, we introduce a novel architecture of multi-task deep neural network coined as Satisfaction and Emotion Neural System for E-learning Improvement (SENSEI) for analyzing users’ satisfaction. The SENSEI combines deep neural network including CNN and BiLSTM algorithms to analyze satisfaction. The experimental results indicate that our proposed model effectively analyzes users’ satisfaction by achieving an average F1-score of $85.51 \%$.