基于IP网络流量数据仓库的KDD实现  

Achieved the Technology of KDD Based on Data Warehouse of IP Network Flows

作　　者：童争雄 刘特[2] 童瑿[3] 

机构地区：[1]上海行知学院,上海200940 [2]上海大学网络中心,上海200073 [3]华东理工大学,海200237

出　　处：《计算机工程与应用》2002年第21期192-195,共4页Computer Engineering and Applications

基　　金：上海市高等学校科学技术发展基金项目(编号:2000A41)

摘　　要：通过对IP网络流量数据仓库进行多表关联检索和决策树模型的数据挖掘,可以从中发现若干有用的知识和相互关联的规则,用于分析流量增长的趋势和寻找IP地址分布与流量大小之间的普遍规律。有助于资源的控制和异常情况的发现。另外,将多表关联算法和决策树挖掘用于星型构架的多维数据集,可以显著地提高数据对象之间的关联性能和数据挖掘的效率。A lot of useful knowledges and relative rules are discovered from the data warehouse of IP network flows which can lead to analyse the trend of data increased and discover the universal rule between the distributing of IP address and IP network flows by using data mining with algorithm of multi-table hash-join and decision-tree.As well as,the IP resource is controlled and the exceptional case is monitored.On the other hand,the performance of relation-ship of data object and efficiency of data mining can be advanced by using the algorithm of multi-table hash-join and decision-tree on MultiDimension Database of Star Schema.

关 键 词：IP网络流量数据仓库 KDD 数据挖掘 决策树 校园网 INTERNET 

分 类 号：TP393.18[自动化与计算机技术—计算机应用技术]