Detection of phishing attacks in Iranian e-banking using a fuzzy–rough hybrid system

• Identifying outstanding phishing features that best fit the Iranian bank websites.
• Extracting a reduct of influential indicators in phishing detection for Iranian e-banking system using rough sets theory.
• Determining critical phishing detection rules and forming a flexible rule base for phishing detection.
• Building a fuzzy–rough hybrid system as a core processing unit of phishing detection applications or web browser add-ons and extensions concentrated on Iranian e-banking.
• Applying the proposed system on Iranian phishing sites and achieving an efficiency of 88%.

Phishing is a method of stealing electronic identity in which social engineering and website forging methods are used in order to mislead users and reveal confidential information having economic value. Destroying the trust between users in business network, phishing has a negative effect on the budding area of e-commerce. Developing countries such as Iran have been recently facing Internet threats like phishing, whose methods, regarding the social differences, may be different from other experiences. Thus, it is necessary to design a suitable detection method for these deceits. The aim of current paper is to provide a phishing detection system to be used in e-banking system in Iran. Identifying the outstanding features of phishing is one of the important prerequisites in design of an accurate system; therefore, in first step, to identify the influential features of phishing that best fit the Iranian bank sites, a list of 28 phishing indicators was prepared. Using feature selection algorithm based on rough sets theory, six main indicators were identified as the most effective factors. The fuzzy expert system was designed using these indicators, afterwards. The results show that the proposed system is able to determine the Iranian phishing sites with a reasonable speed and precision, having an accuracy of 88%.