Multi-Scale Partitions: Application to Spatial and Statistical Databases

We study the impact of scale on data representation from both the modelling and querying points of view. While our starting point was geographical applications, statistical databases also address this problem of data representation at various levels of abstraction. From these requirements, we propose a model which allows: (i) database querying without exact knowledge of the data abstraction level, (ii) the computation of multiple representations of data, one per abstraction level, and (iii) its application to the computation of statistical summaries. The model has been partially implemented with the DBMS O2 by means of tree-structured domains: we give some examples which illustrate the above features.