Hierarchical Reasoning and On-line Analytical Processing

This article presents a hierarchy as an integral part of a data model. Placed in the context of a multidimensional database, this enables summarisation operators such as the CUBE to be extended with hierarchical paths for each dimension. For example, temporal aggregation could be composed of year, month, week, and daily intervals, whereas spatial aggregation could be based on a hierarchy of multi-scale grids, or arbitrary polygons (e.g. enumeration districts and post-code regions). The introduction of functions capable of generating hierarchies for spatial classification and aggregation can reduce the complexity of many empirical studies, such as those investigating the effects of ecological fallacies and the modifiable areal unit problem (MAUP). This lends a more structured methodology to the task, and makes it easier to investigate different parameters in the analysis.