Discovery of spatial links between individuals on the Geospatial Semantic Web

In 2002, Egenhofer described his conceptual vision of the future Geospatial Semantic Web [1]. As a contribution towards the concretization of this approach, we propose here a spatial ontology that can be used to attach spatial characteristics to Web resources, coupled with a spatial reasoner, called ONTOAST. Our work is applied to the field of semantic analysis, a new search paradigm for the Semantic Web, that answers queries like ″Is instance x in any way connected to instance y″. However, semantic analysis has been defined for RDF(S) graphs only, and has, in this context, a limited power of expression. We propose here the adaptation of this search paradigm to OWL 2 ontologies, with the objective of gaining more expressivity and activating more powerful inferences. We also investigate several ways of exploiting spatial information in the process of semantic association discovery. The purpose is to infer new and possibly interesting ontology paths between target instances, based on spatial relations existing between individuals. Then we show how spatial contexts are handled in order to limit the search space to a given spatial region.