Logical design of vague spatial data warehouses

Abstract

Fast, flexible and multidimensional ways for spatial data analysis are provided by tools that query a spatial data warehouse (SDW), which mainly stores crisp spatial data as vector (e.g. political boundaries) and their descriptive attributes (conventional data). On the other hand, spatial vagueness is a spatial data imperfection that hinders distinguishing an object shape from its neighborhood, since the concepts used to describe the spatial information are not defined precisely. As a result, it is not possible to be sure if the parts of a spatial object belong completely to it or partially to it or do not belong to it. Although vague spatial data warehouses (vague SDWs) store both vague spatial data and crisp spatial data, and therefore allow multidimensional and spatial analysis of business scores regarding spatial vagueness, the design and use of vague SDWs are still in infancy. Conceptual modeling of vague SDW according to the cube metaphor is provided by the VSCube conceptual model, proposed by this candidate and this advisor, which defines attribute types, hierarchies and their categories, dimensions, measures, fact, vague spatial fact, data cube and lattice of cuboids to manipulate conventional data, non-geometric spatial data, crisp spatial data and, mainly, vague spatial data. Currently, the logical design of the vague SDW is based on the star schema with fact and dimension tables implemented under the relational model. However, some issues remain opened, especially regarding the logical design of vague SDW: (I) integrity constraints to enable the design of vague SDW under the relational model were not yet specified;(II) attributes of distinct data types and in different levels of granularity may be related through a hierarchy and demand inter level integrity constraints, which were not yet proposed for vague SDWs; (III) the identification of the allowed topological relationships among instances of non-adjacent levels in a hierarchy is not automatic; and (IV) mapping rules between the conceptual model and logical design do not exist. These issues impose severe limitations to the adoption of the vague SDW to address real problems in academy and industry. This research project aims at tackling these issues. Firstly, an investigation will be held to assess the feasibility of reusing and extending the concepts proposed in the VSCube conceptual model to enable the logical design of the vague SDW. Secondly, integrity constraints for the logical design of vague SDW will be defined. Thirdly, an algorithm will be developed to enable an automatic identification for the allowed topological relationships among instances of non-adjacent levels in a hierarchy. Finally, mapping rules between the conceptual model and logical design of vague SDW will be specified. As a result, this research project will provide original contributions to enable representation, storage, manipulation and query of vague spatial data in SDWs, focusing on database logical design. (AU)