Wrapper-Based Feature Selection for Detecting the Lexical Phishing Websites Using Ensemble Learning Algorithms

Abstract

Phishing is one of the most serious issues now-a-days, and many internet users are falling prey to it. Website phishing is a major threat that focuses on developing spoofed sites as a legal ones. Phishers develop cloned websites and distribute the uniform resource locator (URLs) to a large number of people in the form of e-mail, short message service, or through social media. In the current scenario, phishing is the topmost cyber threat/cyber-attack that intrudes into the system to steal or capture sensitive information from the target. Machine learning methods, an important branch of artificial intelligence, are used to detect many critical problems. This chapter investigates the lexical features of website URLs to detect the phishing URL using wrapper-based feature selection on ensemble learning technique. To evaluate the model developed, the dataset from Mendeley repository is taken. The highest level of accuracy for the phishing websites was reached using bagging classifier with 95% accuracy compared with boosting algorithm.