基于Netflow的流量分类方法研究

Traffic classification based on netflow

针对Netflow提供的流量信息有限的问题,在Netflow的基本信息基础上构建更丰富的特征空间,通过机器学方法(决策树、朴素Bayes方法和Bayes网络)研究了Netflow用于流量分类的可行性。实验结果表明,决策树方法在Netflow数据上具有良好的分类效果;同时结合Netflow的广泛性,提出的方法具有良好的实用意义和推广价值。

Due to the limited traffic information provided by Netflow,it is not considered as a suitable data set for traffic classification traditionally.We construct a richer feature space based on Netflow,and use machine learning methods（the decision tree,Navie Bayes and Bayes network）to explore the traffic classification.The experimental results show that the decision tree built on Netflow dataset has better precision than other two methods,and reinforce our suggestion that Netflow is fully appropriate for classification.