A Predictive Model for Phishing Attacks on Mobile Intelligent Agent Systems

Phishing attacks continue to evolve with time and the attackers are always ahead of the existing mitigation plans. The more technology advances, the more sophisticated phishing attacks become. As more people rely on technology to conduct their online businesses, the need for secure Mobile Intelligent Agents (MIAs) systems becomes higher. MIAs are software that execute functions or duties such as managing electronic mail, gathering, and sending out information on behalf of an online user. MIAs play a vital role in carrying out electronic commerce activities whether from business-to-customer, customer-to-customer, or business-to-business. In this paper, we use Decision Tree (DT) to accurately predict the status of a Uniform Resource Locator (URL) and flag those predicted to be phishing. The DT algorithm extract URL features such as the domain name, URL length, host name, page ranking, etc., and feeds them into the model which then makes status prediction. The extracted features are then filtered into the model to improve its future predictions. The proposed model has a prediction accuracy of 91.3% and it is also cost effective as well as easy to use. The other benefit of the proposed solution is its capability to flag phishing URL and prevent a user from launching the web browser thereby protecting their personal information from being stolen by criminals.