Using an Incomplete Data Cube as a Summary Data Sieve

An incomplete data cube is a multidimensional hierarchy of aggregate values in which regions of the hierarchy, and the source data from which those regions are derived, are missing. We model an incomplete cube as a collection of complete sub-cubes called cubettes. Each cubette is deened using a precise, simple speciication. The set of cubette speciications concisely characterises the information content of the cube. We discuss how these simple speciications can be used to generate a parser to sieve data from an underlying data source and populate an incomplete cube.