基于关联挖掘算法的网络数据可追踪共享仿真

Network Data Tracking and Sharing Simulation Based on Association Mining Algorithm

当前方法在实行数据挖掘过程中存在由于产生较大的异常波动导致数据质量受到损害影响数据使用率的问题,现提出基于关联挖掘算法的网络数据可追踪共享仿真方法。利用关联挖掘算法和蚁群算法对网络数据特征实施聚类,从而形成异构网络数据集;然后提取数据集中数据的关联关系,并构建关联矩阵,再采用对称非负矩阵分解处理关联矩阵,形成数据划分指示矩阵;将划分指示矩阵作为关联矩阵三分解的输入信息,以此来实现网络数据可追踪的最佳共享。测试结果表明:当控制波动系数、最小支持度和可信度分别为6、0.3和0.5时,可最大程度降低数据挖掘产生的异常波动,此时数据挖掘的质量较好。实验发现,上述方法的共享数据质量较高、数据之间关联强度较大、数据共享可利用率高。

At present,the data quality is reduced in the process of data mining due to large abnormal fluctua-tions,which affects the data utilization rate.Therefore,this paper proposed a method to simulate network data track-ing and sharing based on association mining algorithm.Firstly,the association mining algorithm and ant colony algo-rithm were adopted to cluster the characteristics of network data,thus forming heterogeneous network data sets.Then,the correlation relationship between data in the data sets was extracted,and the correlation matrix was constructed.Secondly,the symmetric nonnegative matrix was used to decompose the correlation matrix and thus to form the parti-tion indication matrix.On this basis,the matrix was used as the input information of the triangular decomposition of the matrix to realize the best sharing of traceable network data.Test results showed that the quality of data mining was ideal when the control fluctuation coefficient,minimum support and reliability were 6,0.3 and 0.5 respectively,and the abnormal fluctuation caused by data mining could be minimized.The experiment found that the quality of shared data was improved by the proposed method.And the association strength between data was large.In addition,the data sharing allowed for higher availability.