基于用户反馈的混合型垃圾邮件过滤方法

发布时间：2019-03-07 19:12

【摘要】：针对目前垃圾邮件过滤技术仅依赖单一邮件特征实施邮件分类、对邮件特征变化的适应性较差等局限,提出一种基于用户反馈的混合型垃圾邮件过滤方法。以用户社会网络关系为基础,借助用户反馈机制分别实现对基于内容与基于身份标识的邮件分类知识的动态更新;在此基础上采用贝叶斯模型,实现邮件的内容特征与发件人身份标识特征在邮件分类中的有机结合。实验结果表明,与传统的过滤方法比较,所提方法在邮件特征动态变化的环境下能够获得更好的邮件分类效果,邮件分类的总体召回率、查准率、精确率均能达到90%以上。所提方法能够在保证邮件分类性能的同时,有效提高邮件分类对邮件特征变化的适应性,是已有垃圾邮件过滤技术的重要补充。
[Abstract]:A hybrid spam filtering method based on user feedback is proposed in view of the limitation that spam filtering technology only relies on a single e-mail feature to implement e-mail classification and has poor adaptability to the change of E-mail characteristics. Based on the user's social network relationship, the dynamic updating of the content-based and identity-based e-mail classification knowledge is realized by means of the user feedback mechanism. On this basis, the Bayesian model is used to realize the organic combination of the content features of mail and the identity of sender in the classification of e-mail. The experimental results show that compared with the traditional filtering method, the proposed method can achieve better mail classification effect under the environment of dynamic change of mail characteristics. The overall recall rate, precision rate and accuracy of mail classification can all reach more than 90%. The proposed method can not only guarantee the performance of email classification, but also effectively improve the adaptability of email classification to the change of email characteristics. It is an important supplement to the existing spam filtering technology.
【作者单位】：深圳信息职业技术学院计算机学院;南开大学信息技术科学学院;
【基金】：广东省自然科学基金资助项目(S2011040006119) 广东省科技计划项目(2011B040300034)
【分类号】：TP18;TP393.098

【参考文献】