潜在底层网络可产生破坏性的数据挖掘仿真

Simulation of Destructive Data Mining in Potential Underlying Network

随着互连网的不断发展,网络安全威胁的种类越来越多,严重影响了用户正常的使用。在多层网络中,底层网络的数据几乎处于休眠状态,其构成的威胁性的直接属性特征极不明显,无法与其它入侵数据形成可识别的活跃关联特征。传统的网络数据挖掘技术,无法从根本上找到潜在底层网络可产生破坏性的关联规则,无法构建关联检测模型,检测准确率较低。针对这种问题,将数据挖掘技术中关联规则Apriori算法引入到网络数据挖掘中,利用原有Apriori算法的智能性、协作性、互操作性等良好特性,融入网络规则库子系统根据供给的规则集合,进行潜在底层网络可产生破坏性网络数据挖掘,并构建一种基于改进Apriori算法网络数据挖掘算法,用经典的数据集KDDCUP 99对改进算法进行实验,通过改进前后算法各自的检测率和误报率实验数据对比,表明改进算法提高了底层网络威胁数据检测的效率、准确率。

Aiming at the simulation of destructive data mining, the association rules of Apriori algorithm in the data mining technology was introduced into the network data mining. The good characteristics of original Apriori algo- rithm such as intelligent, collaborative, interoperability and others were integrated into the subsystem of network rules library. According to the provided rule sets, the destructive network data mining was performed in the potential un- derlying network. And a network data mining algorithm based on improved Apriori algorithm was built up. The clas- sic data set KDDCUP 99 was utilized to test the improved algorithm. And by comparing the detection rate of experi- mental data with the false rate of the algorithms before and after optimization separately, it can be learnt that the effi- ciency and accuracy of the improved algorithm for detecting threat data in underlying network are enhanced greatly.