基于AdaBoost-Bayes算法的URL分类方法  

URL Classification Method Based on Ada Boost and Bayes Algorithm

作　　者：张腾飞[1] 张谦[1] 刘嘉勇[1] 

机构地区：[1]四川大学电子信息学院,四川成都610065

出　　处：《信息网络安全》2017年第3期66-71,共6页Netinfo Security

基　　金：国家自然科学基金[61377018]

摘　　要：为实现根据HTTP协议数据流对用户的行为进行分析,需要快速区分出用户访问资源的URL。为此文章提出一种结合规则过滤和机器学习算法的方法,用于快速识别用户访问的URL。首先将解析后的数据包根据URL后缀过滤掉资源加载类的数据包,然后根据浏览器UserAgent的特有字段和在浏览器上访问网页的特性识别出浏览器UserAgent, 最后通过基于AdaBoost和Bayse算法训练好的分类器识别出用户访问URL。实验结果表明,本文方法能够在局域网数据流中高效、准确的识别出用户访问的URL。In order to realize the anatysis of the behavior of the data stream from the HTTP protocol, the user needs to identify the URL. In 侃s paper,a new method based on rule filtering and machine learning algorithm is proposed to quickly identify users to access URL. Firstly, the analytical data packets according to the URL suffix filtered load resources packet Secondly, according to the unique browser user agent field and in the browser access identifying characteristic of the web browser user agent. Finalty, the AdaBoost and Bayes algorithm to train a good sub category recognition user access URL based on. Experimental results show that the method can efficiently and accurately identify the user access UI^ ]b the local area network data stream.

关 键 词：规则过滤 机器学习算法 URL分类 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TP393.09[自动化与计算机技术—控制科学与工程]