基于贝叶斯增量分类的邮件过滤研究

随着internet的快速发展,垃圾邮件泛滥成灾.面对垃圾邮件日益严重的现状,提出了贝叶斯邮件过滤模型并讨论了贝叶斯分类方法在垃圾邮件过滤中的应用.针对难以获得大量有类别标签的邮件训练集问题,利用贝叶斯具有增量学习特征,分析并提出了基于小规模训练集的贝叶斯增量邮件过滤方法,通过最小化当前邮件分类器的分类损失,来选择有利于提高分类器性能的邮件加入训练集.实验结果表明,该方法是切实可行的并具有良好的效果.

Research of Mail Filtering Based on Bayesian Incremental Classification

With the rapid development of internet,spam is widely spreaded.Facing the increasingly serious situation of junk mails.The Bayes mail filtration model is proposed,and the application of Bayesian classification method based on the filtration of junk mail is discussed.In view of difficulty to acquire the category label in great quantities of training regulations problem and Bayes feature of incremental study.An incremental bayesian classification is presented based on small training set.Through minimizing the...