Enhancing Phishing Website Detection Using Particle Swarm Optimization and Feature Selection Techniques

In recent years, Web phishing attacks have evolved, eroding customer trust in online services. Traditional blacklist-based approaches struggle to detect sophisticated phishing websites, including newly deployed ones. Researchers are turning to machine learning techniques for early detection but facing limitations in feature selection and parameter tuning. This paper presents an innovative approach that combines Particle Swarm Optimization (PSO) with feature selection techniques, including correlation and mutual information, and tree-based feature selection. The goal is to accurately differentiate between legitimate and phishing websites by identifying relevant features. A comprehensive dataset of diverse website features is used, and correlation and mutual information methods are used to assess feature importance. A tree-based feature selection algorithm refines the feature set, and PSO optimizes the parameters by exploring the feature space. Thorough experimentation is needed to fine-tune features and parameters for accurate detection. Experimental results show the superiority of the proposed approach, eliminating irrelevant features and improving efficiency and classification performance through PSO optimization. Integrating PSO with feature selection provides a robust framework, addressing tuning challenges and improving accuracy.