Domain-specific ontology merging for the semantic Web

Natural language understanding is needed to intelligently handle the large volumes of information that are processed by computers. Most of the data handled by computers is text; the data exists in emails, on Web pages, and in databases. The data is semi-structured; it is not rigorously, unambiguously organized, and constrained. Ontologies may help with analyzing and understanding text. Ontologies provide a capability to represent objects, concepts and other entities and the relationships between them. Ontologies may be used as a tool for finding possible meanings of words in text, and meanings of text in general. We need to be able to merge ontologies from different, sometimes unrelated, sources. Ontologies may be inconsistent, incomplete or imprecise. We consider how to potentially merge two domain-specific ontologies. We assume that both ontologies contain concepts and instances, and both ontologies are large. We suggest to use the information available on the Web using search engines, as well as other methods, such as lexical, semantic and heuristics, to merge two ontologies.