Spear-phishing attack detection using artificial intelligence

• Main Author: Rajkumaradevan, Sanglidevan
• Format: Final Year Project / Dissertation / Thesis
• Published: 2024
• Subjects: T Technology (General); TD Environmental technology. Sanitary engineering
• Online Access: http://eprints.utar.edu.my/6912/1/fyp_CN_2024_RS.pdf; http://eprints.utar.edu.my/6912/

This project focuses on developing machine learning-based applications to enhance cybersecurity, specifically the Spear Phishing Attack Detection (S.P.A.D) system and the Email/SMS Classifier. The goal is to mitigate phishing and spam threats by using advanced algorithms to detect malicious URLs and classify messages effectively. The Spear Phishing Attack Detection system employs models such as Logistic Regression, Random Forest, and ensemble methods to identify and block phishing URLs. It provides real-time feedback on website safety, offering a proactive defense against spear phishing attacks. Extensive testing confirmed the system's accuracy in correctly classifying phishing and legitimate URLs. The Email/SMS Classifier uses models like Naive Bayes, Support Vector Machines, and Random Forest to classify messages as spam or legitimate. The system integrates text preprocessing techniques to enhance classification accuracy and was tested with real-world datasets, demonstrating effective spam detection. Both applications underwent thorough functional, performance, and accuracy testing. Metrics such as precision, recall, and F1 score were used to evaluate effectiveness. The systems were also tested for performance and scalability to handle large data volumes without sacrificing speed or accuracy. The project also explores the characteristics of spear phishing and spam, offering insights into attackers' evolving tactics. These findings inform the development of stronger cybersecurity defenses. Recommendations for future work include refining the models, expanding datasets, and continuously updating systems to adapt to new threats. By integrating these applications into broader security frameworks, their impact could be further enhanced. In summary, this project successfully demonstrates how machine learning can be used to detect and prevent spear phishing and spam, offering innovative solutions to enhance cybersecurity for individuals and organizations.