Incorporando dados espaciais vagos em data warehouses geográficos: a proposta do tipo abstrato de dados vaguegeometry

Abstract

A data warehouse is a solution for organizing and storing multidimensional data related to decision-making processes in companies, generating a historical, highly voluminous, subject-oriented and nonvolatile database. A geographic data warehouse (GDW) additio¬nally to the conventional data storage (i.e. numeric and alphanumeric data), stores spatial data as attributes in dimension tables or as measures in fact tables, storing data represented by geometries. Points, lines and polygons are examples of spatial data types. While spatial data currently stored in GDWs are crisp, i.e., they have exact location in the space, strict interiors and well-defined boundaries, geographic applications have required the storage of vague spatial data, which have inaccurate location, inexact interiors or uncertain bounda¬ries. This Master s research aims at incorporating vague spatial data to GDWs. More speci¬fically, we propose and implement a new abstract data type (ADT) called VagueGeometry to represent vague spatial data in the Spatial Database Management System (SDBMS) Post- greSQL/PostGIS. The proposal of the ADT VagueGeometry encompasses the issue of phy¬sical storage for vague spatial data, which are complex and can have several disjoint parts. It also focuses on definitions of operations to handle vague spatial objects, such as vague topological predicates and its operators. Experimental evaluations were conducted in order to assess the performance of the ADT VagueGeometry in comparison to available solutions, such as implementation of vague topological predicates utilizing existing operations of the PostGIS. The proposed ADT VagueGeometry shown reductions in query processing with vague topological predicates from 81.63% to 90.34% in spatial databases and reductions from 92.46% a 95.20% in GDW environments. This Master s project, therefore, advances in the state of art in GDWs to study this gap in the literature. Additionally, fuzzy models to represent vague spatial data was also studied, and as a result, a preliminary proposal of a ADT, called as FuzzyGeometry, was also developed.