Phishing Website Detection Using a Hybrid Approach Based on Support Vector Machine and Ant Colony Optimization

The Internet has become the most useable tool for people worldwide to communicate and transact business deals, transform money and many other daily activities that require the exchange of sensitive information. As a result, it has become a major platform for cybercrime. In addition, phishing is one of the most common threats faced by internet users. A great number of legitimate and phishing websites come across to the life every day. Therefore, effective methods must be found to deal with the detection and prevention of different types of phishing. Machine learning (ML) techniques have become very popular in phishing website detection by predicting whether a webpage is phishing or legitimate based on datasets obtained from previously extracted features from many other web pages. Thus, this study aims to introduce a phishing website detection approach that utilizes Support Vector Machine (SVM) combined with the Ant Colony Optimization (ACO) algorithm. In addition, the Deep Belief Network (DBN) is also considered to select the best features from the extracted features. The collected dataset contains 12,000 samples. The phishing Uniform Resource Locators (URLs) were retrieved from the PhishTank website, whereas the legitimate URLs were collected from the open datasets of the University of New Brunswick. The performance of the proposed approach was evaluated using accuracy, precision, recall and F1-score metrics. The proposed approach achieved an accuracy of 97.54% surpassing the traditional SVM model by a percentage of 9.58%, and achieved better results on all metrics used.