An Instance Based Methodology for Merging Domain Ontology

Several systems have been proposed to merge ontologies for heterogeneous application domains. They give primacy to concepts and roles. In this paper, we describe an algorithm to merge two or more ontologies constructed from a common set of corpus for an application domain by different groups of ontology experts. We give primacy to instances and values to discover mismatch. Our algorithm makes distinction between roles and attributes of concepts and gives primacy to named instances and attribute values to discover mismatch and to resolve them. We use bottom-up approach and equality relationship between values and names to discover mismatch in ontologies. Names included for consideration are instance name, attribute name and concept name. Our algorithm is demonstrated with two examples.