Case-based content navigation

This paper describes a document retrieval system called CAIRN that uses a c ase-based reasoning set using a large lexicon to automatically generate a c ase index to that document set. The index is used by a case-based retrieval engine to find documents. The retrieval engine is tolerant of noisy natura l language queries. CAIRN also supports failure-driven learning of importan t concepts during its use and thus can significantly improve its retrieval accuracy over time. The limitations of this system are discussed. (C) 1998 Elsevier Science B.V. All rights reserved.