Abstract

ARREDONDO, Raúl  and  MUNOZ, Luis. Repairing Data Warehouses with semantic meaning. Rev cuba cienc informat [online]. 2017, vol.11, n.2, pp.73-86. ISSN 2227-1899.

ABSTRACT A Data Warehouse (DW) is a data repository that is modeled using the multidimensional model, which structures the information according to dimensions and facts. A dimension is an abstract concept that brings together data that share a common semantic meaning, modeled by category hierarchies, which contain elements. A dimension is strict if all category each element has a unique ancestor in each upper and covering category if each element has at least one ancestor each higher category. A dimension can become inconsistent with respect to their strictness constraint and covering integrity after update operations and when this happens the dimension must be repaired. A repair is a new dimension that satisfies the set of strictness and covering constraints and that is obtained inserting and deleting arcs between elements of the categories. If the dimension becomes inconsistent after a single operation reclassification of items it is possible to compute a repair, in polynomial time, but this repair does not guarantee the semantics of the data. In this article different kind of restrictions that guide the repair process such as priority and safe constraints, new heuristics and a algorithmic approach that allow to find a repair with a correct semantic meaning, however, it is possible not to find a repair that guarantees the semantics of the data.

Keywords : Data Warehouse; inconsistency; stricness constraints; covering constraints; heuristics.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )