The design of valid multidimensional star schemas assisted by repair solutions

Multidimensional schemas are used to model data warehouse systems, a special type of large databases dedicated for decision support. Several approaches have been proposed in this domain to ensure the satisfaction of decision makers needs and the accuracy of the generated schemas. This paper presents an approach for the validation of multidimensional star schema assisted by repair solutions. Our approach aims to assist designers by detecting constraint violations and proposing repair solutions based on a number of error-based rules formalized in Prolog. Its efficiency stems from using both linguistic similarity computation and a set of heuristics developed by bottom-up design methods to produce warnings about semantic constraint violations and their impact on the quality of the analysis results.