互联网下多元属性特征恶意停靠域名检测仿真

摘要

对互联网中的用户域名进行安全性检测,在提高用户使用安全性方面具有重要意义.由于互联网的开放性及多样性,使得用户域名形式多样化,传统检测方法主要通过用户域名特征对入侵进行检测,未考虑用户域名形式多样化对用户域名安全性的影响,导致检测准确率、召回率均低的问题.提出基于并行化随机森林法的用户域名保护安全性检测方法,该方法通过对待测用户域名进行定义,并根据经验将用户域名特性划分为四类特征向量,利用域名白名单对域名数据进行预处理,引入并行化随机森林法,结合训练分类器实现用户域名的分类,并对分类器分类结果进行测试,获取用户域名安全性检测结果.仿真结果表明,采用改进方法可提高检测用户域名安全性正确率.%It is of great significance to improve the user's security by detecting the user domain name in the internet.Due to the openness of Internet and diversity,the user domain name form is diversified.Traditional detection method detects intrusion mainly through the features of user domain,without considering the impact on the diversification of user domain user domain security,resulting in low detection accuracy rate and recall rate.The paperj proposes a concurrent user domain name protection safety detection method based on random forest method.This method defined the user domain to be detected,and divided the characteristics of the user name domain into four types of feature vectors according to experiences,Domain name data were pre processed with the domain name list,and parallel random forest method was introduced.Combining the training classifier to achieve classification of the user names,and the classification results were tested to obtain the user domain security detection results.The simulation results show that the improved method can improve the security,accuracy and recall rate of the user's domain name.