Résumé : |
(auteur) An increasing number of geospatial applications require information which is scattered in several independent geographic information systems. One of the main objectives of geographic information infrastructure, GII, is to provide a political, institutional, economic, and technical platform to share information. The focus of this thesis is on the technical aspects of the GII. The thesis aims at providing a mechanism to share information seamlessly among distributed, heterogeneous, geospatial information systems. Sharing information may improve decision making and reduce the cost of data collection. In general, retrieving information from distributed databases involves two steps. In the first step users search for relevant information resources in a network of information providers. In the second step users request data from the information resource. With regard to the first step, a model to search for the relevant information resources is presented. The model is called the resource discovery model, RDM. It provides a reference model to structure the metadata of the information resources in a tree of interrelated resources. In the second step, interoperability allows communication among heterogeneous, distributed, information systems. Interoperability is the ability of two or more systems to exchange geospatial information and to make mutual use of the information that has been exchanged. The research identifies two perspectives to interoperability; these are the data modeling perspective and the system architecture perspective. In relation to data modeling, three types of heterogeneity arise: syntactic, schematic, and semantic. The semantic heterogeneity occurs due to differences in the definition of classes, the definition of class intension and the geometric description. This set of definitions is called context information. The semantic heterogeneity is the main factor for the schematic and the syntactic heterogeneity. The schematic heterogeneity pertains to the differences in the class hierarchies and the attribute structure of database schemas. The syntactic heterogeneity occurs due to the differences in the constructs used to model relationships among classes and attributes, object geometry, and topologic relationships. To provide interoperability among different GIS applications, it is necessary to resolve the semantic, schematic, and syntactic heterogeneity. In this thesis, a model for information sharing is presented. The model is called the semantic formal data structure, SFDS. It consists of three layers, in which each layer is intended to resolve a specific type of heterogeneity. The model provides a method for loading semantics that is the context information, into database schemas. The first layer of SFDS is the syntactic layer, in which the formal data structure, FDS, is adopted. The second layer of SFDS is the schematic layer, in which the concept of federated databases is adopted. A reference model for the federated schemas is presented. The implementation of SFDS and RDM is related to the system architecture perspective for interoperability, and is discussed in this thesis. A comparison of the implementation of RDM, which is a clearinghouse, with other implementations has proven that a consistent abstract model is required to maintain and ensure the consistency of the contents of the clearinghouse as well as to improve the results of the search for information resources. The three-layers approach adopted in SFDS has proven adequate to resolve the three types of heterogeneity. The implementation of SFDS, known as the semantic translator, has shown that it should be dedicated to a single application domain, to simplify its practical implementation and maintenance. In this case databases can have several semantic translators installed, each being specific to an application domain. For example, one database may have a semantic translator to exchange road network information, another to exchange soil information, etc. |