垃圾邮件的改进贝叶斯过滤算法

Improved Bayesian filtering algorithm for spam

在研究贝叶斯过滤算法原理和实现方法的基础上,将垃圾邮件的先验概率由常数改进为实际概率,改进了token的选取范围和选取规则,在检测内容上增加url和图片。最后设计了一个基于改进贝叶斯过滤算法的垃圾邮件过滤器。实验结果表明,这种改进的贝叶斯过滤算法在垃圾邮件过滤中有良好的应用效果。

Based on the theory and practice of Bayesian filtering algorithms, a detailed process for improving the algorithm is put forward. Firstly, instead of a constant probability of spam, the actual prior probability is used. Secondly, the selective scope and rules of token are improved. Finally, url and images are added to the detected content. A spare filter based on the improved Bayesian filtering algorithm has been designed. Test results show that the improved Bayesian filtering algorithm works well in practice.