Email Phishing Detection System Using Neural Network
Amina A. Abdullah, Loay Edwar George, J. Mohammed · Research Journal of Information Technology · 2015
One of the internet based identity thefts is called phishing. Email phishing is a way that phishers trick the user to give information. The increasing of the phish attack in the resent years cause a lot of problems; credit card number, user name, password were stolen due to the phish attack. Due to this attack people lose their money, personal information and trust in online business also the attack effects on the companies' reputation. Many companies were losing money up to millions of dollars. In this study a new stem that can quickly detect phishing emails with low false positive rate is introduced. A set of features is proposed to test each coming emails to identify whether it is phish email or not. A feed forward neural network with back propagation training algorithm was adopted to categorize the email samples into phish or ham category. First a set of extracted features vectors from each email, each vector consists of 17 features, had been used to make the categorization decision. Then, a set of smaller feature vectors, each consists of only the 12 best features, was used for classification tests. The results achieved in this study are 99.91 (for 17 features) and 99.95for (12 best features).