Resumo

RAMIREZ, Juan-José; ARREDONDO, Raúl  e  VALLEJOS, Cristian. Compatible Repair in Inconsistent Data Warehouses. Rev cuba cienc informat [online]. 2018, vol.12, n.2, pp. 15-29. ISSN 2227-1899.

A dimension in a data warehouse is an abstract concept that groups data that shares a common semantic meaning. Dimensions are modeled by a hierarchical scheme of categories. A dimension is called strict if each element of each category has a unique ancestor in each higher category, and homogenous if each element of a category has at least one ancestor in each higher category. If a dimension is strict and homogenous, precomputed queries can be used at the lower levels to obtain answers at higher levels. However, when the dimensions are not strict/homogenous their constraints should be known in order to obtain a correct result. In the real world, dimensions may not meet these constraints, and in these cases, it is important to correct these dimensions or find ways to get correct answers to questions posed in inconsistent dimensions. A minimal repair is a new dimension that satisfies the strict and homogenous constraints, and is obtained from the original dimension through a minimum number of changes, which has an NP-hard computational cost. To improve this, the extended dimension is defined and the compatible repair is proposed, which obtains a new consistent dimension, maintaining the essence of the original dimension. The experimentation is focused in this last point, to elaborate and evaluate a program that generates this compatible repair and obtains results in polynomial time.

Palavras-chave : data warehouse; experiments; homogenous constrains; inconsistency; strict constrains.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )