基于案例推理的历史登陆战役地理环境分析  

Case-Based Reasoning for Geo-Feature Indexing and Analyzing of Typical Landing Operation

作　　者：蓝荣钦[1] 杨晓梅[2] 

机构地区：[1]解放军信息工程大学测绘学院,郑州450052 [2]中国科学院资源与环境信息系统国家重点实验室,北京100101

出　　处：《地球信息科学》2004年第3期7-11,共5页Geo-information Science

基　　金：国家863项目(2003AA604040)资助.

摘　　要：基于案例推理是智能决策辅助系统中的一种重要技术。本文结合历史登陆战役地理环境，对登陆战役所涉及的各种地理要素进行了分析，把登陆地域所涉及的领域背景知识表示成概念层次树形式，使数据的分布模式具有更强的可视性。将案例推理技术引用于历史登陆战役地理环境查询分析系统，研究案例库的结构设计、案例的组织和表示、地理环境处置规则的表示等。叙述了案例库的建立方法、属性权重算法和相似性度量算法，以及系统的结构和实现方法。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(K

关 键 词：案例推理 战场地理环境 属性权重 相似性度量 

分 类 号：E835[军事—战术学]