Abstract

In adversarial systems, the performance of a classifier decreases after it is deployed, as the adversary learns to defeat it. Recently, adversarial data mining was introduced as a solution to this, where the classification problem is viewed as a game mechanism between an adversary and an intelligent and adaptive classifier. Over the last years, phishing fraud through malicious email messages has been a serious threat that affects global security and economy, where traditional spam filtering techniques have shown to be ineffective. In this domain, using dynamic games of incomplete information, a game theoretic data mining framework is proposed in order to build an adversary aware classifier for phishing fraud detection. To build the classifier, an online version of the Weighted Margin Support Vector Machines with a game theoretic prior knowledge function is proposed. In this paper, a new content-based feature extraction technique for phishing filtering is described. Experiments show that the proposed classifier is highly competitive compared with previously proposed online classification algorithms in this adversarial environment, and promising results where obtained using traditional machine learning techniques over extracted features. Copyright 2009 ACM.