基于元数据关联特征的交互式数据快速查询

Interactive Data Quick Query Based on Metadata Association Characteristics

大数据时代下数据查询频率逐渐增多,为提升用户搜寻信息准确性与实时性,提出一种基于元数据关联特征的交互式数据快速查询方法。分析大数据交互式流程,在交互式数据预处理阶段定位异常数据,透明化交互过程,支持数据溯源,让用户理解结果与初始数据相对关系,给用户极佳操作体验;运用Map Reduce编程模型完成元数据操作处理,使用单次扫描方式,并行元数据抽样得到精确元数据关联结果;利用相空间重构方式组建高维相空间,代入F.Takens嵌入理论,获取元数据关联差分累积函数特征,并将元数据关联特征矩阵奇异值分解,完成交互式数据快速查询。仿真结果证明,所提方法的交互式数据查询准确率较高,且提升了数据关联查询效率。

A fast query method based on metadata association feature is presented for improving the accuracy and real-time of users’ searching information. The interactive process of big data was analyzed in detail to locate abnormal data, make the interactive process transparent and trace the source of supporting data, so that users can understand the relative relationship between the results and the initial data for supplying users excellent operation experience. Based on the Map Reduce programming model, the operation of metadata was completed. The single scan method was used to parallelize the sampling of metadata, and the accurate results of metadata Association were obtained. According to the reconstruction method of phase space, high dimensional phase space was established, and F. Takens embedding theory was used to obtain the characteristics of metadata association differential cumulative function. Meanwhile, the singular value of the associated characteristic matrix of metadata was decomposed. Eventually, the fast query of interactive data was achieved. The simulation results show that the method has high interactive data query accuracy and improved data association query efficiency.