摘要
基于案例的推理(Case—Based Reasoning)是一种基于记忆，利用过去的案例和经验来解决新问题的一种方法，它可以看作是从一个案例到另一个案例的类比推理。由于CBR具有易学易用，知识获取容易，适应性强等优点，受到人工智能研究者的高度重视，是人工智能领域新兴的一种重要的推理技术。其基本解题步骤可分为五步：1)新问题的描述；2)案例的检索；3)案例方案的调整；4)案例方案的评估；5)案例学习和维护。
本文依次介绍了案例表示的相关知识及一种面向对象的表示方法，案例库组织索引的目标、原则及采用动态聚类进行组织索引的方法，几种典型的案例检索方法，案例调整知识的获取、调整方法的分类及一个转换式调整模型，新案例的评估方法及案例库的学习与维护等。
此外，对于案例检索过程中的两个重要问题，属性权重的设置和属性值间局部相似度的赋值问题作了重点讨论。对于权重设置，先简要介绍了已有的一种基于粗糙集的权重自动学习算法，并对其进行了问题分析和改进，还介绍了一种权向量渐进学习算法，当用户对当前权值不满意时，可采用此算法进行逐步修正。对于局部相似度的賦值问题，可根据属性类型的不同，即数值型属性，无序枚举型属性，有序枚举型属性分別采用不同的方法来度量其相似度值。
本文最后重点讨论CBR技术在农业专家系统中的具体应用。阐述了使用Delphi 7.0程序设计语言开发的“CBR大豆专家系统”的具体实现及从推理效率和准确性方面进行的实验分析，对CBR技术在农业领域中的具体应用作了初步的探索。但其中仍有些技术不太成熟，如案例调整过程的自动化问题，以及CBR与其它技术的进一步结合，如何开发出更加完善的CBR系统是我们需要进一步研究和努力的方向。
Case-Based Reasoning is a memory-based method which utilizes old cases and experience to solve new problem, and it can be seen as analogism from a case to another case. Because GBR has much merit such as easily learning and using, facilitated knowledge acquirement, strong adaptability etc, it gets highly regard of artificial intelligence researcher and is an important reasoning technique rising in artifical intelligence field. Its-basic solving steps is divided into five steps: 1) description of new problem; 2) retrieval of cases; 3) adjustment of scheme; 4) evaluate of scheme; 5) learning and maintenance of case.
This paper in turn introduces correlative knowledge on case description and a Object Oriented representation, the aim, tenet of casebase organizing and index and a organizing and index method using dynamic clustering, several typical case retrieval methods, the acquirement of adjustment knowledge, the classification of adjustment methods and a transform adjustment model, evaluate method of new case and learning and maintenance of casebase etc.
Moreover this paper also stressed discusses two central problem in case retrieval: the setting of property weight and the assignment of local similar degree between property values. For weight setting, first briefly introduces a known weight learning arithmetic based on rough set, and carrying through problem analysis and improvement, also introduces a weight vector gradually learning arithmetic, when user isn't satisfied with current weights, we can use it to revise weights gradually. For local similar degree assignment, we can according to the difference of property types, namely numerical type, disorderly enumeration type and orderly enumeration type, then separately use different method to measure similar degree.
Lastly this paper discusses the application of CBR technique in agricultural expert system. Expatiate on material realization of "CBR bean expert system" developed using Delphi 7.0, and experiment analysis on reasoning efficiency and veracity, doing primary explore for the practical application of CBR technique in agricultural field. But also there are some techniques which aren't mature, such as automatization of case adjustment, and further combination of CBR and other techniques, so how to develop out more perfect CBR system is our future research and struggling direction.
引文
[1] 石纯一，黄昌宁等．人工智能原理．清华大学出版社，1993年10月第1版
[2] 孙即祥等编著．现代模式识别．国防科技大学出版社，2002年1月第1版
[3] Roger Schank. Dynamic Memory: a Theory of Reminding and Learning in Computers and People. Cambridge: Cambridge University Press, 1982
[4] 高莹，钱建华等．基于案例推理对群体决策的支持．管理信息系统，1999；(6)
[5] 史忠植．知识发现．清华大学出版社，2002
[6] Jagielska I, Matthews C, Whitfort T. An investigation into the application of neural networks, fuzzy logic, genetic algorithms, and rough sets to automated knowledge acquisition for classification problems. Neurocomputing, 1999,24:37-54
[7] Azuaje F, et al. Discovering relevance knowledge in data: a growing cell structure approach. IEEE Transactions on systems, man, and cybernetics, 2000, 30(3)
[8] Anand S S, et al. Discovering case knowledge using data mining. In: Proc. Of PAKDD,-98, 1998.25-35
[9] Flinter S, Keane M T. On the automatic generation of case libraries by chunking chess games. In: Proc, of the first Intl. Conf. On Case-based Reasoning, 1995. 421-430
[10] 李宝东．基于范例的学习——模型和方法研究．1991硕士学位论文，中国科学院计算技术研究所
[11] Babaka o, Whar S Y. Case-Based Reasoning and decision support systems. In: IEEE Intl. Conf. on Intelligent Processing Systems. Beijing, 1997. 1532-1536
[12] 汤胤，彭宏，郑启伦．基于XML的多层次案例映射模型．计算机科学2004，31(2)
[13] Rodriguez A F, Vadeera S, Sucar L E. A probabilistic model for case-based reasoning. In: I Smith, B Faltings, eds. Cased-Based Reasoning Research and Development, Proc. Of ICCBR-97,1997. 623-632
[14] Skalak D. Prototype and features selection by sampling and random mutation hill-climbing algorithm. In: Proc. 11th Int. Machine Learning Conf, Morgan Kaufmann, 1994. 293-301
[15] MeSherry D. Demand-driven discovery of adaptation knowledge. In: Intl. Joint Conf. On Artificial Intelligence, Volume 1, 1999. 222-227
[16] Kitano H, Shimazu H, Shibata A. Case-Method: A methodology for building large-scale
case-based systems. In: Proc. of AAAI93, 1993. 303-308
[17] Bull M, Kundt G, Gied L. Discovering of health risks and case-based forecastiong of epidemics in a health surveillance system. In Jan Komorowski and Jan Zytkow editors, Principles of Data Mining and Knowledge Discovery, 1997. 68-77
[18] Sbreese J, Heckerman D. Decision-theoretic case-based reasoning, In: Proc. Of the fifth Intl. Workshop on Artifical Intelligence and Statistics, 1995. 56-63
[19] Aha D W, Chang L W. Cooperative Bayesian and case-based reasoning for solving multiagent planning tasks: [Technical report]. Navy Center for Applied Research in AI, Naval Research Laboratory, Washington, DC, USA, 1996
[20] 张荣梅，胡玉胜，涂序彦．基于范例推理的交通事故智能处理系统．小型微型计算机系统，2001，22(10)
[21] Li Baodong, Shi Zhongzhi. 1992. Case retrieval based on memory network. In: Shi Zhongzhi, ed. Automated Reasoning, North-Holland, 275-285
[22] Ren Haitao, Li Ru etc. Realization Approach Research On CBR Agricultural Expert System. Proceeding of the Second International Symposium on Intelligent Information Technology in Agriculture
[23] 吴良刚，高阳，张金隆．一种面向对象的事例知识表示方法和事例库组织．管理信息系统，1999：(7)：11～14
[24] 倪志伟，蔡庆生．基于知识发现的范例推理系统．计算机科学，2003，30(5)
[25] 李媛．PKRS系统的设计与实现．哈尔滨工业大学工学硕士学位论文
[26] 耿焕同，钱权，蔡庆生．基于聚类策略的一种范例删除模型．计算机科学，2003，30(4)
[27] Wei-Chou Chen, Shian-Shyong Tseng, Lu-Ping Chang etc. A parallelized indexing method for large-scale case-based reasoning. Expert System with Applications, 2002(23): 95-102
[28] 郭茂祖，苏晓红，王亚东，李媛．基于IBL算法的CBR系统中索引与检索机制研究．计算机工程与应用，2001，37(5)：67～69
[29] 唐智勇，韦东鹏，王红胜．基于线索和改进最相邻近法的案例检索．计算机工程与应用，2000，36(4)：78～79
[30] Quinlan J R. 1979. Discovering rules from large collections of examples: A case study. In: Michie D, ed. Expert systems in the Micro Ellectronic Age, Edinburgh University Press
[31] Kraslawski A, Pedrycz W, Nystrom L. Fuzzy neural network as instance generator for case-based reasoning system. Neural Computing&Applications, 1999, 8: 106-113
[32] Yahia M E,et al. Rough neural expert systems. Expert systems with applications, 2000, 18:
87-99
[33] Ralph Bergmann and Wolfgang Wilke. Towards a New Formal Model of Transformational Adaptation in Case-Based Reasoning. ECAI 98. 13th European Conference on Artificial Intelligence, Edited by Henri Prade, Published in 1998 by John Wiley & Sons, Ltd.
[34] Smyth B, Keane M T. Remembering to forget: A Competence-Preserving Case Deletion Policy for Case-Based Reasoning Systems. In: Proc. 14th Intl. Joint Conf. On Aritificial Intelligence, 1995: 337-382
[35] 骆敏舟，周美立．实例推理检索中相似度量方法的研究[J]．合肥工业大学学报(自然科学版)，2001，24(6)：1091～1094
[36] Takao Mohri and Hidehiko Tanaka. An Optimal Weighting Criterion of Case Indexing for Both Numeric and Symbolic Attributes. http: //citeseer.nj.nec.com/akkus96weighting.html
[37] 王国胤，于洪，杨大春．基于条件信息熵的决策表约简．计算机学报，2002，25(7)
[38] 任海涛，李茹．案例特征权重自动学习方法研究．电脑开发与应用，2004，17(3)
[39] 吴昊，杜小鸿，刘学馨．粗糙集和数据预处理[J]．高技术通讯，2000，10增刊
[40] 张文修，吴伟志，梁吉业，李德玉．粗糙集理论与方法[M]．科学出版社，2001
[41] Stanfill.C, Waltz. D. Towards Memory-based Reasoning. Communications of the ACM 29(12).1986:1213-1228
[42] Chunjiang Zhao. Progress of Agricultural Information Technology. International Academic Publishers, 2000
[43] 高技术通讯．2000，10增刊．科学技术部863计划联合办公室，中国科学技术信息研究所
[44] Aynur Akkus, H.Altay Guvenir. Weighting Features in k Nearest Neighbor Classification on Feature Projections. http://citeseer.nj.nec.com/akkus96weighting.html
[45] Hector Munoz-Avila, David W.Aha, etc. SiN: Integrationg Case-based Reasoning with Task Decomposition. http: //citeseer.nj.nec.com/akkus96weighting.html
[46] 杭小树，熊范伦．基于CBR的农作物病虫害预报专家系统．计算机工程与应用，2000，36(10)：161～163
[47] Frank Heister and Wolfgang Wilke. An Architecture for Maintaining Case-Based Reasoning Systems
[48] Andrew R.Golding, Paul S.Rosenbloom. Improving Rule-Based Systems through Case-Based Reasoning. In Proceedings of AAAI, Anaheim, CA, 1991.