ARTICLES
LAN Rongqin, YANG Xiaomei, Zhengzhou
.
2004, 6(3):
7-11.
Based on cognitive psychologic process,Case-Based Reasoning(CBR)is a new method for machineself-learning and solving problem.CBR has been widely applied in many study fields served as an alternativeapproach to rulebased reasoning for solving some complex problems.Especially it has become an importantreasoning technique for intelligent decision support system.In the field of battlefield geographic environment analysis,historical landing operation geographicenvironment is very helpful to current decision-maker.By applying CBR,they can be treated as typical casesand manipulated with an example battle base.In this paper,the important value of geo-feature indexing andanalyzing of typical landing operation is reviewed.The geo-features related to landing operation are explored indetail and then classified with concept hierarchy.The techniques of CBR are applied to the design of geo-feature indexing and analyzing system of typical landing operation.They are discussed from three aspects.The first is Case Base Construction.In our system,a typical landing operation case is formally representedas the following form:Case=(problem,constrained condition,solution),which is the main body consisting ofentities and attributes of the problem.Constrained condition means the domain expert knowledge or rulerequired for constructing cases.The second is Case Indexing.When finding a solution for a new problem,CBRuses similar knowledge and feature indexing to search the most similar case to current problem from case base.This includes three sub-processes:characteristic discrimination,similarity computation,and test and evaluationwith domain expert knowledge,in which similarity computation is most important.Generally,the similarity ofthe case can be divided into four levels,that is,attribute similarity,entity similarity,level similarity,and totalsimilarity,there are many algorithms and equations for these similarity computations.The third is Selecting Indexing Strategy.There are three major case indexing strategies for CBR:NearestNeighbor(KNN),Inductive Indexing and Knowledge Guiding.In our system,since the organization andexpression of the case are based on KNN,the indexing strategy is called weighted indexing based on KNN.Finally,based on the framework of CBR and its key techniques,a case base of typical landing operationshas been built using programming language VB and VC as well as the database management system SQL Serverand Microsoft Access.It has been proved a high performance.