Mouse Dynamics-Based Online Fraud Detection System for Online Education Platforms
Conference Publication ResearchOnline@JCUAbstract
Over the past decade, the online education sector has gained significant popularity causing a major shift from traditional face-to-face classes to online education. This shift has left the education sector with major vulnerabilities. Specifically online fraud, which includes identity theft and online assessment fraud. Conventional methods used for online fraud detection are one-time, invasive, and expensive. Behavioral biometric authentication systems based on mouse dynamics have emerged as a secure and inexpensive alternative to conventional methods due to their ability to continuously track user identity non-invasively. However, there are no studies on mouse behavioral biometric-based continuous user identification, for online fraud detection, in online education platforms. Furthermore, there is also a need to identify a better combination of features that yield satisfactory performance (>80% in accuracy, precision, and recall) for this specific application. Existing research in the field still relies on typical machine learning methods despite major developments made in deep learning. To solve the identified problems, we first identify a unique combination of mouse behavior features and collect application-oriented mouse behavior data. We then investigate the efficacy of the features using both deep learning and typical machine learning methods. Experimental results demonstrate that the identified feature set, termed as user behavior-centric feature set, not only yields satisfactory performance but is reliable and scalable to significantly larger datasets.
Journal
N/A
Publication Name
ICICT 2024: 9th International Congress on Information and Communication Technology
Volume
1003
ISBN/ISSN
978-981-97-3301-9
Edition
N/A
Issue
N/A
Pages Count
13
Location
London, UK
Publisher
Springer Nature
Publisher Url
N/A
Publisher Location
Singapore
Publish Date
N/A
Url
N/A
Date
N/A
EISSN
N/A
DOI
10.1007/978-981-97-3302-6_21