Merge of X-ETL and XCube towards a standard hybrid method for designing data warehouses

Abstract

There is no doubt that the hybrid approach is the best paradigm for designing effective multidimensional schemas. Its strength lies in its ability to combine the top-down and bottom-up approaches, thus exploiting the advantages of both approaches. In this paper, the authors try to identify and analyze the different hybrid methods developed for building data warehouses. The analysis revealed that the existing methods are too complicated and time consuming in the deployment phase. In order to solve this problem, the authors introduced a new hybrid method that is easy to use and saves a huge amount of deployment time. This new method consists of two main steps: the first data driven step allows an analysis of the source models by using the X-ETL method and gives rise to star models. The second requirements driven step performs a semantic analysis of the needs expressed in natural language by using the XCube Assist method. This analysis allows to improve the quality of star models generated by the X-ETL method without the intervention of a designer.