Hardening web browser security configuration using machine learning technique

Abstract

Browser configuration settings play important role such that no or less information of user or user's system will be available to attacker or rogue website. The default browser configuration is often not adequate to stop or minimise information leakage to the attacker. In this paper, a novel model (framework) to bridge the gap between default and recommended configuration is proposed. The framework is developed using machine learning algorithm, as huge set of browser configuration states need to be classified into different security levels. A prototype browser add-on is developed using the framework to assess browser security level and modify it to increase security level if required.