摘要
信息粒度原理是一种从多个角度来精确描述对象的物理学方法。本文将信息粒度的原理应用到垃圾邮件的过滤中,提出了一种基于信息粒度原理的垃圾邮件过滤方法。通过对原始样本空间更精细的划分来实现对邮件类别的更准确描述。本文在Ling-Spam语料库上进行了试验,结果表明,新方法具有较高的分类精度和良好的处理速度。
The principle of information granularity is a physical way which can describe the object much better. We introduced this concept into spam filtering, and a method based on information granularity is proposed to filter spam, and exact description about mail＇s class information can be achieved through fine partition on training mails. Experiment on Ling-Spam corpus shows the effectiveness with respect of classification accuracy and speed.
出处
《信息工程大学学报》
2007年第1期15-17,52,共4页
Journal of Information Engineering University
关键词
信息粒度
垃圾邮件过滤
类重心向量
information granularity
spam filtering
category centroid vector
