A framework of integrated continuous online learner with fuzzy neural network application in learning environment

• Main Author: Sadikan, Siti Fairuz Nurr
• Format: Thesis
• Language: English; English; English
• Published: 2021
• Subjects: QA76 Computer software; TJ212-225 Control engineering systems. Automatic machinery (General)
• Online Access: http://eprints.uthm.edu.my/1868/2/SITI%20FAIRUZ%20NURR%20BINTI%20SADIKAN%20-%20delaration.pdf; http://eprints.uthm.edu.my/1868/1/SITI%20FAIRUZ%20NURR%20BINTI%20SADIKAN%20-%2024p.pdf; http://eprints.uthm.edu.my/1868/3/SITI%20FAIRUZ%20NURR%20BINTI%20SADIKAN%20-%20fulltext.pdf; http://eprints.uthm.edu.my/1868/
• Institution: Universiti Tun Hussein Onn Malaysia

Online learning has become popular among university due to its flexibility and adaptability. This technology offers the capability of learning, anytime and anywhere, based on student preferences. However, the general method of Personal Identification Number (pin) to verify the user does not guarantee a person's identity, because other people can use it. Even occasional visitors and users tend to pass their tokens or share their passwords with their colleagues to make their work easier. In all scenarios, online assessments such as quiz, test and examination are conducted without face-to-face supervisions. This situation potentially leads the students to find help from their peers or other sources to get high scores. This research addressed the issue related to the online assessment. The main objective of this work was to propose the use of the Online Learner Verification Framework (OLVF). This proposed solution utilizes the keystroke analysis and activity-based authentication for the online learner authentication. Besides, a fuzzy neural network approach was used to train and validate the learners’ identity to predict their cheating tendency. In addition, challenge questions are also generated randomly by the system, based on user profiling. An online learning system was designed specifically in this study to simulate an original online learning assessment. It is expected to contribute to the field of security, where dynamic profile queries are asked, based on the previous history extracted from the online learning system. The proposed framework was validated using experimental datasets from an online learning system, elearning2u.com. The results obtained showed that, the proposed framework is able to overcome problems associated with the existing methods, thus improving the security level of the current online learning systems. Among others, appropriate questions and answers for the system-generated challenging questions are such that, invalid users will find them very difficult to guess. As a result, valid users too will find it difficult to provide their usernames and passwords to a third party or to ask others to answer online assessments for them. The results obtained proved this framework to be the safest method to be used, if implemented in the current online learning system.