一种基于用户会话的异常检测方法

随着网络技术的发展,人们对网络的依赖性越来越强,但同时网络攻击给网络用户造成了严重的信息泄露和巨大的经济损失。如何从浩瀚的用户访问信息中发现对网站具有恶意攻击行为的用户就成为了Web服务管理者亟需解决的重要问题。对Web服务日志的深入分析后,发现攻击访问用户与正常访问用户在访问Web服务时形成的日志记录具有不同的特征。通过特征提取并且进行必要假设后,利用朴素贝叶斯分类算法构建异常检测分类模型,取得了较好的检测效果。

An Anomaly Detection Method Based on Session

With the development of network technology,the dependence of the network is more and more strong,but simultaneously the network attack cause serious leak of information and huge economic losses. How to find out the attacker from vast user access information is an important issue needs to be solved for Web service administrator. After the deep analysis of Web service log,find that there are dif-ferent in character between abnormal and normal access users. By feature extracting and making some necessary assumptions,build anom-aly detection model using Na？ve Bayesian classifier with a good detection effect.