Spatio-temporal Multi-granularity: Modelling and Implementation Challenges

Abstract. Multiple spatial and temporal granularities are essential to extract significant knowledge from datasets at different levels of detail: they enable to zoom-in and zoom-out a dataset, enhancing the data modelling flexibility and being instrumental to boost the analysis of information. Implementing granularities poses several interesting problems. Specifically, in this paper we analyse the issues involved by enhancing a data model and a query language with spatio-temporal multi-granularity, and we figure out efficacious solutions to address all of them. In our exposition, we investigate proper representa-tions for the spatial and the temporal domains; then we conceive an appropriate design for granules and granularities, and for multi-granular values. In particular, mutual relationships among granular-ities and how they affect granularities design is discussed according to their influence on data access and considering the application of multi-granular conversions. Afterward, we dedicated to the design of multi-granular spatio-temporal conversions, discussing the multiple ways in which they affect data usability and envisaging how the design of a multi-granular model and query language may guarantee such an fundamental property, reducing uncertainty on the represented values, combining concepts like topologically consistent transformation, probability distributions, invertibility and quasi-invertibility properties. In our discussion, we are influenced on our previous work on multi-granularity. Especially