关系数据库中分布式大数据集成冲突消解仿真

Distributed Big Data Integration Conflict Resolution Simulation in Relational Database

针对当前方法在进行冲突消解时存在消解耗时较长、消解成功率较低的问题,提出一种基于BP神经网络的关系数据库分布式大数据集成冲突消解方法,利用相似度度量方法提取关系数据库中分布式大数据集成过程数据属性特征在语义上的冲突特征,包括字符类型属性值的数据、数值类型属性值的数据、布尔类型属性值的数据,还有区间值类型属性值的数据四种;在相似度计算基础上,实现不同属性值类型数据的冲突特征提取,将这些冲突特征输入到训练好的BP神经网络模型中,判断关系数据库中分布式大数据集成过程是否存在冲突,并对存在的冲突进行消解。仿真对比测试结果证明,所提方法能够实现关系数据库中分布式大数据集成过程的冲突消解,而且具有耗时低、成功率高的优点。

In order to solve the long time consumption and low success rate in the conflict resolution,this paper puts forward a method of distributed big data integration conflict resolution in relational database based on BP neural network.At first,the similarity measure method was used to extract the semantic conflict feature of data attribute feature of distributed big data integration process in relational database,which included the data of character type attribute value,the data of numeric type attribute value,the data of Boolean type attribute value and the data of interval type attribute value.On the basis of similarity computation,the conflict characteristics of data with different attribute types were extracted.After that,these conflict characteristics were input to the trained BP neural network model,thus we could judge whether there were conflicts during distributed big data integration in relational database,and then we resolved existing conflicts.According to simulation results,the proposed method can realize the conflict resolution in the relational database during distributed big data integration.Meanwhile,this method has low time consumption and high success rate.