The Research Of IDSS Based On Inductive Learning And Case-Based Reasoning

Decision Support System is integrated with Artificial Intelligence to constitute Intelligence Decision Support System (IDSS). IDSS can effectively solve the problem of how qualitative analysis is combined with quantitative analysis and those structured and non-structured problems, extend its scope which it can deal with, and improve the decision ability. IDSS has been widely applied to some fields, such as preventing flood, supporting e-business. Inductive Learning and Case-based Reasoning(CBR) are both key techniques of machine learning which has fully been developed, and they have been widely researched and applied. However with the application of IDSS, the problem of "the bottleneck of acquiring knowledge" has gradually come out and hindered its development in some extent. Fortunately the two key techniques can applied to IDSS to effectively solve the problem mentioned above, so as to effectively and efficiently support the decision.This thesis firstly discusses the principle, architecture construction, function, research situation and existing problems of IDSS. The application of Machine Learning is researched to solve the problem of "the bottleneck of acquiring knowledge" that exists in the study of IDSS. And it introduces some algorithms of decision tree learning such as ID3, C4.5 and feature subset selection of Inductive Learning. Then it also introduces some basic techniques of CBR such as case representation, case retrieval, case revisal, case base maintenance.According to these theories mentioned above, the thesis raises some new application of Inductive Learning in CBR. Some algorithms of Decision tree learning, especially C4.5, are applied to case retrieval, case revisal and case base maintenance.And also several algorithms of feature subset selection, especially the algorithm which is based on the information entropy and Laplace error evaluation functions and able to deal with noisy data, are applied to the reduction of attributes of case in the case base.Finally, combined with the decision of preventing flood, the thesis has studied the application of Inductive Learning and CBR in IDSS and designed the prototype of an integrated IDSS based on Inductive Learning and CBR, presented integrated architecture and key techniques, so that it can make a wide space to research IDSS.