Anwar, Aamir (2024) Emotion-Aware Online Learning: A Hybrid Sentiment Analysis-based Model to Augment Online Learning Pedagogies using Artificial Intelligence Algorithm. Doctoral thesis, University of West London.

PDF Aamir Anwar - Final thesis (clean) (Aug 2024).pdf - Published Version Restricted to Repository staff only until 1 August 2027. Available under License Creative Commons Attribution Non-commercial. Download (13MB)

Abstract

The rapid growth and expansion of online learning, especially during the Coronavirus Disease
2019 (COVID-19) pandemic, have highlighted its significance across various sectors, includ�ing traditional education, corporate training, and workshops. However, challenges such as
student engagement and satisfaction during learning sessions continue to affect its effective�ness. Students’ engagement and emotions play an essential role in learning, directly impact�ing their satisfaction with the content, the platform, and the instructor’s teaching methods.
Effective student engagement can be achieved through continuous prompting and gamifica�tion, providing timely feedback during learning sessions, and creating balanced groups of
students in collaborative learning settings. Emotional recognition can be achieved using fea�ture extraction from images and videos, sentiment analysis of students’ textual utterances,
physiological sensors, and speech or voice recognition.
This thesis addresses two major challenges in online learning: achieving balanced hetero�geneous groups in collaborative learning to improve student engagement and accurately
predicting students’ emotions in real-time, as it requires accurate recognition and classifica�tion of multiple emotions in real-time from different modalities.
This thesis introduces a novel activity-based technique for Dynamic Group Formation (DGF)
to address the group formation challenge. This technique automatically swaps students into
different groups based on their learning styles and knowledge levels to ensure balanced het�erogeneous groups. These balanced groups are then used in the Intelligent Tutor-Supported
Collaborative Learning System (ITSCL), an online platform designed for collaborative learn�ing, to improve the learning process and increase educational gains. Additionally, the pro�posed technique is validated through user experience experiments to evaluate its practical
application in real-world scenarios
Another significant contribution of this thesis is the development of a meta-emotional model.
This model utilises data from Student Utterances (SUs) and facial gestures to improve stu�dent satisfaction and engagement collected during 10 online learning sessions. The proposed
meta-emotional model employs a transfer learning-based Bi-directional Long Short-Term
Memory (Bi-LSTM) deep learning model to classify SUs into emotional categories: engaged,
bored, confused, frustrated, and neutral. Furthermore, a Convolutional Neural Network
(CNN)-based modified MobileNet model is used to classify facial gestures into these emo�tional categories.
An intelligent online learning platform (Intelli-Student) has been developed to analyse the
effectiveness of the meta-emotional model. This platform integrates meta-emotional mod�els (SUs and facial gestures) with an Intelligent Tutoring System (ITS). ITS main purpose
is to intervene during student inactivity and prompting in the online learning session. The
Intelli-Student platform recognises students’ emotions from SUs and facial gestures and clas�sifies them into academic emotion classes. The system’s emotion detection modules provide
real-time feedback to the instructor about students’ understanding and engagement during
learning sessions. Moreover, the system provides overall class-level and individual student
engagement levels to the instructor during and at the end of each session. This information
offers insights into student learning experiences and satisfaction during learning sessions,
which helps to improve online pedagogies, learning content, and student engagement. By
providing timely and precise feedback, the system enhances the adaptability and responsive�ness of online learning environments, ensuring a more personalised and effective educational experience for each student.

Item Type:	Thesis (Doctoral)
Identifier:	10.36828/thesis/12329
Subjects:	Computing > Intelligent systems
Depositing User:	Marc Forster
Date Deposited:	20 Aug 2024 08:39
Last Modified:	04 Nov 2024 11:30
URI:	https://repository.uwl.ac.uk/id/eprint/12329

Actions (login required)

View Item