改进多维关联规则算法在多源异构数据挖掘中的应用

Application of Improved Multidimensional Association Rule Algorithm in Multi-source Heterogeneous Data Mining

针对传统多源异构数据挖掘方法存在的数据处理速度慢、加速比和分类精度低的问题,研究了改进多维关联规则算法在多源异构数据挖掘中的应用。利用改进多维关联规则算法分析数据之间的关联性,建立多源异构数据特征单元集合,根据数据权重与属性之间的相似度计算多源异构数据可信度,完成多源异构数据的预处理。计算多源异构数据的优先权值,并对数据进行排序处理,采用全局极小理论计算多源异构数据的适应度,以此搭建多源异构数据对象模型。以多源异构数据对象模型为依据,搜索出具有最大先验概率的数据划分策略,根据多源异构数据优化的目标函数设计了多源异构数据挖掘算法,实现了多源异构数据挖掘。实验结果表明,本文方法可以提高多源异构数据的处理速度,在加速比和分类精度上也具有一定的优势。

Aiming at the problems of slow data processing speed,low acceleration ratio and low classification accuracy of traditional multi-source heterogeneous data mining methods,the application of improved multi-dimensional as⁃sociation rule algorithm in multi-source heterogeneous data mining was studied.The improved multidimensional association rule algorithm was used to analyze the association between data,establish the feature unit set of multi source heterogeneous data,calculate the reliability of multi-source heterogeneous data according to the similarity between data weight and attribute,and complete the preprocessing of multi-source heterogeneous data.The priority value of multi-source heterogeneous data was calculated,and the data was sorted.The global minimum theory was used to calculate the fitness of multi-source heterogeneous data,so as to build the multi-source heterogeneous data object model.Based on the multi-source heterogeneous data object model,the data partition strategy with the maximum prior probability was searched out.According to the objective function of multi-source heterogeneous data optimization,the multi-source heterogeneous data mining algorithm was designed to realize multi-source heterogeneous data mining.Experimental results show that the proposed method can improve the processing speed of multi-source heterogeneous data,and has certain advantages in acceleration ratio and classification accuracy.