Abstract: As the electronic commerce and on-line trade expand, phishing is among the hardest network problems. This paper presents a novel approach for intelligent phishing web detection based on Na?ve Bayes and imbalanced support vector machine. As given a web, its URL features are first analyzed, and classified by Na?ve Bayes. When the web’s legality is still suspicious, its web pages are then analyzed, and classified by imbalanced support vector machine. Compared to the existing methods, experimental results show that this approach can achieve the high detection accuracy and the lower detection time.

Key words: phishing, Na?ve Bayes, imbalanced support vector machine, classifier

摘要： 随着电子商务和在线交易的不断发展，钓鱼网站已成为目前最难处理的网络安全难题之一。提出了一种基于贝叶斯和不平衡支持向量机的钓鱼网站检测方法，首先提取待检测网站的URL特征，采用改进贝叶斯方法进行分类检测，如果不能明确分类，则提取该网站的页面特征，并采用不平衡支持向量机方法进行分类检测。实验结果表明，与现有方法相比，方法所需的检测时间少且能达到较高的检测准确度。

关键词: 钓鱼网站, 贝叶斯, 不平衡支持向量机, 分类