ENHANCING THE DETECTION OF PHISHING WEBSITE USING SUPPORT VECTOR MACHINE AND SIMULATED USER BEHAVIOUR IN MATLAB ENVIRONMENT

Abstract

Background: However, the incorporation of behavioural features introduced a novel method of detecting phishing websites which offers valuable insights into user interaction patterns. Objective: The increasing daily presence on the internet has exposed users to several cybercrimes; among them, phishing has emerged as the most prevalent and damaging attack. Methods: The introduction of behavioural features, such as mouse movement intensity and click frequency, in combination with static attributes like URL length and the number of dots, created a more robust detection model. The research findings showed that while static features are significant in detecting phishing sites to a maximum of 97.54% accuracy, the incorporation of behavioural features complemented insights which enabled the developed model to detect phishing attempts in real-time with 100% accuracy. The simulated real-time web traffic showed that whenever the model is applied in real life settings, good accuracy, precision, recall, and zero false positive rates for both the baseline and integrated models were achieved. Results: The results from this research adds to the wealth of knowledge in the field of cybersecurity as it enhances phishing detection systems and creates opportunities for future exploration of machine learning algorithms that combine dynamic user behaviors to achieve better solutions to cyber threats. Conclusion: This research is on enhanced phishing website detection system which combines Support Vector Machines (SVMs) with simulated user behaviour in a MATLAB environment.