基于用户交互的钓鱼邮件检测机制及其实现

发布时间：2018-03-11 09:24

本文选题：钓鱼邮件　切入点：检测　出处：《计算机工程与应用》2017年15期 　论文类型：期刊论文

【摘要】：针对国内外钓鱼邮件研究没有注重个体差异以及缺乏细粒度用户交互反馈的情况,提出一种集理论研究与实际应用相结合的框架,构建一种可视化邮件异常特征模型,设计一个chrome插件,使用分类器对邮件进行检测,将邮件异常特征详细信息和判定情况实时报告给用户。用户查看邮件后结合实际情况与分类器判定情况进行对照检查,出现判定不一致时对相应特征信息进行校正,将校正后的特征向量反馈给分类器进行修正,并将相应的特征信息组加入黑名单或白名单。实验证明,框架能根据个体反馈修正分类器,有效提高钓鱼邮件检出率和用户对钓鱼邮件细粒度感知体验。
[Abstract]:According to the research at home and abroad fishing mail is not focused on individual differences and lack of fine-grained user feedback, put forward the framework of a set of combination of theoretical research and practical application, to build a visual mail anomaly model, the design of a chrome plug-in, using the classifier to detect the mail, mail information with abnormal characteristics and determine the real-time report to the user. The user view the message according to the actual situation and the judgment of classifier check, determining inconsistencies of the corresponding feature information is corrected, the feature vector corrected feedback to the classifier was modified, and the characteristic information of the corresponding group added to the blacklist or whitelist. Experiments show that according to the framework of individual feedback correction classifier, effectively improve the detection rate of phishing emails and user experience of phishing emails fine-grained perception.

【作者单位】：空天信息安全与可信计算教育部重点实验室;武汉大学计算机学院;
【基金】：国家自然科学基金(No.U1636107,No.61202387,No.61373168)
【分类号】：TP393.098

【相似文献】