Data Consolidation Based on Structure Granulation and Matrix Computation

DOI：10.11896/j.issn.1002-137X.2017.09.049

 作者 单位 E-mail 闫林 河南师范大学计算机与信息工程学院 新乡453007 高伟 河南师范大学计算机与信息工程学院 新乡453007 gaoweihsd@foxmail.com 闫硕 北京交通大学计算机与信息技术学院 北京100044

为了研究数据合并问题,并使合并数据保持合并前的数据之间的关联关系,对各类数据信息给予了结构化的表示,对应产生了由数据集和加权关系组合构成的加权关联结构；进而通过数据集的合并粒化集,完成了加权关联结构向加权粒化结构的转换,使数据集中的数据依据粒化信息得到了合并,并保持或汇集了合并前的数据之间的关联信息,由此形成了数据合并的结构粒化方法。在此基础上,构建了加权关联矩阵和加权粒化矩阵,分别作为加权关联结构和加权粒化结构的矩阵表示。经中间变换和目标变换的矩阵计算,实现了加权关联矩阵向加权粒化矩阵的变换,产生了与结构粒化等价的矩阵变换方法,形成了程序设计的算法基础。

In order to study the problem of data consolidation,and make the consolidation data keep the associated information before consolidation,a structure composed of a data set together with a weighted relation was constructed,which refers to a weighted association structure.It demonstrates a structured representation of various data information.Then by using a granulation set,the weighted association structure is transformed into the weighted granulation structure.This makes the data in a granule to be consolidated according to the granulation information.At the same time,the consolidation data maintain or gather the associated information before consolidation,which leads to a structural granulation method of data consolidation.On this basis,two matrices were introduced,called the weighted association matrix and weighted granulation matrix which are matrix representations of the weighted association structure and weighted granulation structure respectively.Moreover,by matrix computations of elementary transformation and target transformation,the weighted association matrix is transformed into the weighted granulation matrix.This forms an equiva-lent way of data consolidation,and can be taken as the basis of programming algorithm.