MDAG-Cubing: A Reduced Star-Cubing Approach

Joubert de Castro LimaCelso Massaki Hirata

In this paper, we extend the Star-Cubing approach by introducing a new hybrid dimension-based approach to efficiently compute full or iceberg cubes with simple or complex measures. This new approach, named Multidimensional Direct Acyclic Graph Cubing (MDAG-Cubing), introduces the notion of external and internal nodes to reduce the cube representation without loss of generality. The reduced representation enables improvements in memory consumption and data structure traversals, since it has smaller height, fewer nodes and fewer branches. We implement two MDAG algorithms that run, on average, 25-50% faster than the Star-Cubing algorithm and consume 70-90% less memory to represent the same data cube.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: