Abstract

Phishing is an illicit attempt to obtain sensitive information from users, such as usernames, passwords, and credit card details. Attackers often use disguised URLs to deceive web users and steal private information. Attackers commonly employ social engineering tactics like email and text messaging to facilitate phishing attacks. Active learning techniques, including neural networks, can help identify and prevent phishing scams, though these methods have certain limitations. This study provides a comprehensive review of various Artificial Neural Network approaches that incorporate decision trees and optimal feature selection techniques to address these challenges. We propose a newly developed feature assessment index that we can combine with an optimal feature selection method. Techniques like decision trees, paired with local search methods, are used to prune unnecessary features, enhancing the effectiveness of phishing detection. The study also examines various social engineering attacks, particularly phishing websites, which often serve as entry points for numerous fraudulent activities.

Keywords: Phishing attacks, machine learning, artificial intelligence, social engineering attacks, cybercrime, scenario-based methods.