Technology of separate formation of multidimensional data with lists of measure values
Sergey V. Zykin, S.V. Mosin, Andrey Poluyanov · 2015
The aim of this paper is to study Hypercube building from relational database using separate specification of dimensions and measures. We don't require any transformation of the initial Database schema, even to the hierarchical model; that allows us to apply this technology without modification of existing databases and their applications. Also we don't force the dimensions to functionally determine measures. In this case, providing data integrity becomes more complicated, but in exchange we get the ability to use non-key attributes as dimensions. Furthermore, we can store a list of homogeneous values in a single cell. We present a formal definition of source and target data representation. The hypercube building algorithm is based on a sequence of intermodel mappings. Hypercube integrity is achieved by using an intermediate representation, called Table of Joins (TJ). This approach is the development of traditional OLAP technology.