Learning Case Classiication for Improving Case-based Reasoning
- Daniel Matuschek, Klaus P. Jantke
This paper deenes a class of problem domains called`knowl-edge mediation' problems where knowledge and information are exchanged between agents operating in two diierent conceptual worlds. Case-based reasoning systems for knowledge mediation will contain cases which are instances of translation between two points of view. These systems require facilities for the automatic adaptation of cases since the user will be unable to fully comprehend the parts of the case which refer to the foreign domain. We described a prototype knowledge mediation system which mediates between the points of view of business managers and engineers and show how`structural similarity' has provided a powerful mechanism for adapting retrieved cases.