Static, Dynamic, and Hybrid Feature Dataset for Lightweight Detection of Malicious JavaScript in Web Applications

This dataset contains static, dynamic, and hybrid feature representations of benign and malicious JavaScript samples collected for research in web security and malware detection. The dataset was created as part of a postgraduate dissertation titled “Lightweight AI-Driven Hybrid Detection of Malicious JavaScript in Web Applications Using a Browser Extension.”

The static dataset includes lexical, statistical, and structural features extracted from JavaScript source code without execution. The dynamic dataset captures lightweight runtime-observable intent features, designed to be feasible within browser extension environments where full execution control is not available. The hybrid dataset combines both static and dynamic features to improve robustness against obfuscation and delayed execution behavior.

Raw JavaScript samples are not included due to security, legal, and licensing considerations. The dataset is intended for academic and research purposes, including comparative evaluation of machine learning models for malicious JavaScript detection and the development of lightweight browser-based security solutions