Merging Expressive Spatial Ontologies Using Formal Concept Analysis with Uncertainty Considerations

Abstract

In this chapter, we present a solution to the problem of merging structures that represent the conceptual layer of some information systems. The kind of structures we are studying correspond to expressive ontologies formalized in Description Logics. The proposed approach creates a merged ontology which captures the knowledge of a set of source ontologies. A main constraint to our solution consists in the fact that instances associated to the concepts of the source ontologies are available. Then it is possible to apply the techniques associated to Formal Concept Analysis. The main contributions of this work are (i) enabling the creation of concepts not originally in the source ontologies, (ii) providing a definition to these concepts in terms of elements of the source ontologies and (iii) handling the creation of merged ontologies based on the uncertainties encountered at the object and alignment levels. This approach is particularly useful in domains where ontologies are intensively exploited. This is typically the case for spatial information where for instance, the nature of land parcels can be characterized by a geographical ontology.