Reconciliation of Disparate Earth Observation Thematic Data through Semantics Driven Middleware

There is a growing demand for digital databases of topographic and thematic information for a multitude of applications in environmental management, and also in data integration and efficient updating of other spatially oriented data. These thematic datasets are highly heterogeneous in syntax, structure and semantics as they are produced and provided by a variety of agencies having different definitions, standards and applications of the data. In this paper we focus on the semantic heterogeneity in thematic information sources, as it has been widely recognized that semantic conflicts are responsible for the most serious data heterogeneity problems hindering the efficient interoperability between heterogeneous information sources. In particular, we focus on the semantic heterogeneities present in the land cover classification schemes corresponding to the North America Land cover characterization data. We propose a framework (Semantics Enabled Thematic data Integration (SETI)) that describes in depth the methodology involved in the reconciliation of such semantic conflicts by adopting the emerging semantic web technologies. Ontologies were developed for the classification schemes and a shared ontology approach for integration of the application level Ontologies is described. A semantics driven Graphical User Interface (GUI) has been developed in Java that implements the SETI framework for the integration of disparate thematic data sources. We employ Description Logics (DL) based reasoning on the terminological knowledgebase developed for the land cover characterization which enables querying and retrieval that goes beyond just keyword based searches.