On indexing in case based reasoning applied to pre-transportation decision making for hazardous waste handling

Abstract

Case Based Reasoning is an Artificial Intelligence technique that utilises the concept of reasoning from a stored knowledge base reflecting prior experience. It offers a number of advantages, such as learning and validation of solutions, over traditional rule based reasoning techniques in unstructured application domains. This paper concentrates on indexing techniques tested with a view to applying a Case-Based Reasoning approach to decision support in the pre-transportation decision making phase of hazardous waste handling. The paper first makes a brief introduction to the field of Case-Based Reasoning. A short overview of the major research areas as well as the advantages and disadvantages is given. As the major focus of the paper, theoretical and practical issues surrounding the indexing problem in Case-Based Reasoning are then covered in greater detail. A short outline of the current decision making process at Waste-Tech (SA) is given, together with a proposed method for incorporating a Case-Based decision support system into the process. The paper then discusses the indexing techniques designed and tested for this particular problem area. Two platforms were used for implementation and testing. Firstly, ESTEEM, a commercial Case-Based Reasoning development tool, was used to implement and test certain techniques as a control. Secondly, techniques were implemented and tested in Delphi. The discussion of techniques is followed by a description of methods used for testing them. Testing results are presented, and a discussion is included, comparing the techniques to each other and to the results of other Case-Based reasoning research.