垃圾邮件优化过滤方法的研究与仿真  

Research and Simulation on Optimization of Spam Filtering

作　　者：邵叶秦[1] 

机构地区：[1]南通大学现代教育技术中心,江苏南通226019

出　　处：《计算机仿真》2013年第12期265-268,共4页Computer Simulation

基　　金：国家自然科学基金项目(61171132);江苏省自然科学基金项目(BK2010280);南通市应用研究计划项目课题(BK2011003;BK2012034)

摘　　要：研究网络通信垃圾邮件过滤优化问题,由于存在随机性,传统的垃圾邮件过滤方法均以内容过滤为主,未考虑其它因素,邮件内容过多会造成确认特征过多,而内容过少则会影响确认的准确性,未全面考虑其它可识别特在因素,具有检测效率低和误差率高的缺陷。提出一种基于Kalmann滤波器和主元特征相似判断模型的垃圾邮件过滤方法,提取邮件的结构主元特征,采用Kalmann滤波器的权值更新模型调整特征项的权重,如果出现新信息,则运算全部特征相权重和,并将其转化成概率值,若概率值高于设置的阀值,通过主元相似判断模型分析待识别邮件,准确过滤相应的垃圾邮件。实验结果表明,改进方法能够具有较高的垃圾邮件识别效率和精确度,具有较强的应用价值。In this paper, the optimization problem of spam filtering in network communication was researched. An improved spam filtering method was proposed based on Kalmann filter and the evaluation model of similar principal characteristics to extract the main element structural characteristics of the message. Firstly, using the weight of Kalmann filter, the model was updated and the weight of feature item was adjusted. If there is new information, the sum of all weights of feature item was calculated, and it was converted into probability values ; if the probability value is higher than the threshold, the evaluation model of similar principal characteristics was used to identify the mail and accurately filter the spam. Experimental results show that the improved method has high spare detection efficiency and accuracy as well as strong application value.

关 键 词：垃圾邮件过滤 主元相似判断 权重 概率值 滤波器 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]