Shallow representations of information (e.g., semantic graphs and networks) are good at providing structure for large amounts of information, but such representations are known to be highly ambiguous. Our project will start by investigating how to augment shallow representations with richer information, expressed as ontologies in formal representation languages, such as OWL. This will help avoid ambiguity and may also expose infidelities and inconsistencies in the shallow information. Since information is likely to come from multiple sources, the project will consider using techniques such as Distributed Description Logics and other ontology modularization mechanisms to better-understand how semantic graphs may import information from other graphs. The project also proposes to point the way toward effective reasoners that combine large amounts of both shallow and ontology information, and will study dynamic change in shallow information and ontologies.
Document last modified on August 31, 2007.