The omnipresence of case-based reasoning in science and application

A surprisingly large number of research disciplines have contributed towards the development of knowledge on lazy problem solving which is characterized by its stor age of ground cases and its demand driven response to queries Case based reasoning CBR is an alternative increasingly popular approach for designing expert systems that implements this approach This paper lists pointers to some contributions in some related disciplines that o er insights for CBR research We then outline a small number of Navy applications based on this approach that demonstrate its breadth of applicability Finally we list a few successful and failed attempts to apply CBR and list some predictions on the future roles of CBR in applications Case Based Reasoning Case based reasoning CBR is a multi disciplinary subject that focuses on the reuse of experiences i e cases It is di cult to nd consensus on more detailed de nitions of CBR because it means di erent things to di erent groups of people For example consider its interpretation by the following three groups Cognitive Scientists CBR is a plausible high level model for cognitive processing Kolodner a Arti cial Intelligence Researchers CBR is a computational paradigm for problem solving Aamodt Plaza Expert System Practitioners CBR is a design model for expert systems that can be used in either stand alone or embedded architectures Watson To introduce CBR requires identifying it in a particular context For example variants of Aamodt and Plaza s problem solving cycle are frequently used when introducing CBR to AI re searchers Figure displays a view of ve top level steps that combined input a problem descrip tion and output a proposed solution Retrieve Given a problem retrieve a set of stored cases e g hproblem solution outcomei triplets whose problems are judged as similar Reuse Apply one or more solutions from these retrieved cases perhaps by combining them with each other or with other knowledge sources Revise Adapt the retrieved solution s as needed in an attempt to solve the new problem Review Evaluate the outcome s when applying the constructed solution to the current problem If the outcome is not acceptable then the solution will require further revision Retain Consider adding the new triplet of problem revised solution and outcome to the library as a new case Case Library New Problem 1. Retrieve Input