关键词: 钓鱼邮件, 检测, 可视化, 异常特征, 互动

Abstract: Aiming at the situation of phishing email researches worldwide not paying attention to individual differences and lacking of fine-grained user feedback, this paper proposes a framework which combines theoretical research and practical application. A visual model of email’s abnormal features is constructed, and a chrome plug-in which uses the classifier to detect the email and reports the details of abnormal features and the judgment to the user in real time is also designed. After the user checking the email, he/she examines the classifier’s judgment in accordance with the actual situation. Then, the corresponding feature information is corrected when the judgment is inconsistent. Therefore, the corrected feature vector is fed back to the classifier for correction, and the corresponding feature information is added to the blacklist or white list. In conclusion, the experiment shows that the framework can modify the classifier according to the individual feedback, meanwhile, improve the detection rate of phishing email and the user’s fine-grained perception of phishing email effectively.

Key words: phishing email, detection, visualization, abnormal feature, interaction