]> Forcing all communicating agents to share a common ontology is infeasible. A group of people with similar interests usually has its own organizational schemes for documents. This organization may be in the form of an ontology. Different agents may define very different ontologies, and the semantics for the same terms may be very different in their ontologies. A mapping from one agent's ontology to another agent's ontology is required to facilitate communication between agents.

This project develops a tool for automatic mapping between concepts of two ontologies. We use a probabilistic approach for ontology mapping by using explicit information in the form of documents assigned to individual concepts. This approach combines DAML+OIL (for ontology specification), IR text classification techniques (for raw similarity scores between concepts from each ontology) and Bayesian reasoning (for mapping generation). User specified landmarks are used to improve accuracy and efficiency.

]]> 2001-09-01T00:00:00-05:00 2002-08-01T00:00:00-05:00 2002-07-07T00:00:00-05:00