Malicious web content detection

Abstract

A "malicious web page" poses a significant threat to internet security by containing harmful content capable of exploiting client-side computer systems. Cybercriminals leverage these pages for various nefarious activities such as phishing, drive-by downloads, and spamming. Detecting such vulnerabilities has become increasingly challenging due to the continual evolution of attack techniques, compounded by a lack of awareness among users regarding potential exploits. The advent of dynamic HTML further empowers attackers, enabling them to compromise computer systems with greater efficiency. Traditional signature-based antivirus solutions often struggle to identify disguised malicious HTML codes effectively. To address this gap, the project proposes a malicious web page detection system employing the Random Forest algorithm. By systematically analysing the characteristics of a web page, our approach demonstrates resilience to code obfuscations and accurately determines the malicious nature of a webpage. Experimental results validate the efficacy of our method in enhancing internet security against evolving threats.