基于多项式朴素贝叶斯算法的垃圾邮件过滤器的设计与实现  

作　　者：李腾飞 

机构地区：[1]河南大学计算机与信息工程学院,河南开封475000

出　　处：《科技资讯》2018年第33期1-2,4,共3页Science & Technology Information

摘　　要：基于概率的朴素贝叶斯分类器因其算法复杂度低、分类精度高而被广泛应用于垃圾邮件过滤领域。该文在对传统朴素贝叶斯分类器进行分析的同时,结合垃圾邮件过滤的特性,设计并实现了基于多项式朴素贝叶斯算法的垃圾邮件过滤器。该过滤器引入拉普拉斯平滑因子降低合法邮件被误判为垃圾邮件的概率,得到了较好的分类效果。实验结果验证了方法的有效性。Probability-based naive bayes classif iers are widely used in spam f iltering because of their low algorithm complexity and high classif ication accuracy. In this paper, the traditional naive bayes classif ier is analyzed, and combined with the characteristics of spam f iltering, a spam f ilter based on polynomial naive bayes algorithm is designed and implemented. The f ilter introduces a Laplacian smoothing factor to reduce the probability that a legitimate mail is misjudged as spam, and a better classif ication effect is obtained. The experimental results verify the effectiveness of the method.

关 键 词：垃圾邮件分类 多项式朴素贝叶斯 网格搜索 平滑因子 

分 类 号：TP393.098[自动化与计算机技术—计算机应用技术]