Efficient Android Phishing Detection Based on Improved Naïve Bayes Algorithm

With the rapid development of the mobile Internet, phishing attacks are becoming more common on mobile phones. In order to effectively detect phishing attacks on Android platforms, this paper proposes an improved framework based on the revised Naive Bayes algorithm. Under this framework, the K-means algorithm is used to supplement missing values of attributes to get the complete datasets. The probability is enlarged to resolve the problem of low biased estimation of the Bayesian algorithm. Weights of different attributes are evaluated to avoid neglecting the relationship among them to improve the accuracy of phishing website detection. The probability ratio of phishing websites to legitimate websites is adjusted to further improve the correct rate of detection. Experimental results have demonstrated that the proposed framework can effectively detect the phishing attacks with relatively small time cost.