The research subject of this PhD thesis is information integration from heterogeneous and distributed information sources. The presented research provides realization of the model for integration of heterogeneous geospatial information sources based on ontologies. Proposed solution is a part of ongoing work on the GeoNis framework. GeoNis framework provides a data-interchange infrastructure for a large number of geoinformation systems that communicate with each other using standard protocols. Every geoinformation system that participates in the data-interchange process has to provide a translator/wrapper component to enable information flow to and from the GeoNis infrastructure. Proposed solution deals with the problem of creating an appropriate translator/wrapper component whenever a new geoinformation system is incorporated into the GeoNis infrastructure.
The GeoNis framework is based on a hybrid ontology approach for data integration. For this reason, a very important subject in presented research on semantic data integration is the creation of mapping between a geospatial information source and its local ontology. This PhD thesis propose a methodology and a tool for the automatic extraction of local ontologies and appropriate mappings from a local geospatial database. OWL2RDB mapping language is proposed for creation of an intermediate layer between a relational database and the OWL ontology. This intermediate layer contains rules (expressed in the OWL2RDB language) for mapping between the structural elements of a relational database and the concepts of OWL ontologies. A system that uses the OWL2RDB intermediate layer to create classes that can handle ontology instances stored in relational databases is also presented in this thesis.
Based od proposed model for integration of geospatial information sources, a prototype of a tool for the automatic generation of translator/wrapper components in the GeoNis interoperability environment is implemented. The implemented prototype was used to evaluate proposed method and to practice it with real data. Conducted analysis pointed out advantages that are achieved with proposed methodology, but also helped in identifying some shortcomings with possible solutions to overcome them.