The design and implementation of two-dimentional spatial data types and their integration into a relational database management system

Abstract

The term spatial data signifies all geometric objects of different dimensions and the relations that could bind these objects together. Spatial data is, in general, divided into two parts: crisp spatial data and vague spatial data. Crisp spatial data has determinate coordinates and boundaries, whereas vague spatial data has indeterminate ones. The demand for spatial data support in database management systems is growing every day, yet not all commercial products provide spatial data support natively, and those that do merely provide it partially. Since database management systems are at the heart of most data-oriented applications, generic support for non--traditional data types (such as spatial data types) is required from such systems. The main goal of this thesis is to analyze spatial data in detail and to propose a method for storing and manipulating such data. Both crisp and vague spatial data were analyzed, and their usage and major aspects were highlighted. We also studied existing support for spatial data in commercial database management systems and pointed out the existing lack of functionality in those systems. We propose several ways of storing crisp spatial data based on some Open GIS Consortium (OGC) standards and on manipulating it by means of computational geometry. We also propose ways to represent and manipulate vague spatial data. These propositions were based on sampling techniques, which are used in the filed on GIS. In order to prove our concepts, we designed and implemented a spatial extension package (SPEX) and integrated it in a commercial relational database management system. We also devised the necessary visual components and a software prototype to test our package. We conclude our thesis by enforcing the idea of the "common object store," which is a DBMS capable of storing and manipulating all sorts of data (relational data, objects, multimedia, spatial, etc.).