Detecting Internet Phishing Attacks Using Data Mining Methods

Abstract

Abstract-Nowadays, the high rate of internet usage among cell phone users has caused many commercial and financial services to be provided through the internet. Despite the fact that internet has provided a functional platform for financial transactions of the users, it can also become challenging and dangerous. Information theft or phishing is a security challenge which is usually carried out by sending spoofed emails and spams. In this type of attacks, attackers usually try to earn the trust of the users by sending deceptive emails in order to direct them towards certain websites which contain spoofed pages for capturing user information. Phishing attacks based on spams are one of the most significant barriers for expanding online financial activities which can cause great losses for financial and credit institutions annually. Assessing the behavior of hackers in sending spam emails and carrying out phishing attacks shows that these attacks follow a particular pattern which is not discernable at the first glance. Discovering the hidden patterns of phishing attacks can be effective in developing software systems for protecting against this type of attacks. Data mining possesses a number of methods for extracting useful information from different unorganized data. This study tries to use a set of data mining tools for analyzing the data from these attacks in order to identify and detect the useful patterns of these attacks. The overall results show that compared to other methods, neural network is more accurate and more sensitive in detecting such attacks.