中文摘要：

提出一种基于邮件头信息的三支决策垃圾邮件过滤方法。该方法使用一种新的属性重要度度量方法,并用该度量方法将邮件头信息属性依据重要度大小进行排序,然后按属性重要度的大小顺序对邮件计算贝叶斯概率并进行三支决策。当信息较少以致不足以决策时,按属性重要度大小顺序增加新的属性信息以帮助进一步的决策,直到得到最后的邮件分类。对比实验结果表明,该方法是合理且有效的。

英文摘要：

A method of three-way decision spam filtering was proposed in this paper based on the head information of E-mail. The head information is sorted by a new measurement of attribute significance. Bayesian probability based on the most significant attributes is computed to do the actions of three-way decisions. When the information is not enough to make decisions,more attribute information is added to the computing of Bayesian probability until the final decisions are made. The results of comparative experiments show that the new method is reasonable and effective.