A Neural Network Method of Similar Case Retrieval in Case-based Reasoning

In order to overcome the handicap of obtaining case attribute weights in classical nearest-neighbor approach to retrieve similar cases in case-based reasoning(CBR),the similarity of precondition and the similarity of conclusion are defined,and a neural network approach is used to learn the case attribute weights.In this method,the case attribute weights are obtained automatically from the case library.Using the method,the case attribute weights are attained from cases.The similar cases are retrieved by nearestneighbor classification with the learned case attribute weights.Finally,the approach is used in photodynamic therapy(PDT) Port Wine Stain(PWS) medicinal database.