| As a new pattern of reasoning, case-based reasoning (CBR) has been developed in the recent two decades. It has aroused intensive attention of the researchers in the artificial intelligence domain. It has been a hot topic and cutting-edge in the area of artificial intelligence and machine learning. The CBR system solves the current problem by retrieving the similar problems from the case base. It overcomes the low efficiency in dealing with the problems outside the bound and matching collision and other defects in the traditional knowledge process system.In the CBR system, the case base is becoming increasingly larger with the incremental learning which results in the decline of case retrieval efficiency and its weaker performance. So the case retrieval is the bottleneck for the system.In this thesis, the history of CBR is first reviewed, followed by the introduction of current research of CBR as well as its characteristics, application fields and theoretical basis. Then the reasoning technologies of CBR and similarity computation are discussed in detail. This thesis also introduces current research and related technology of data mining, and emphasizes a clustering algorithm which is one of important algorithms of data mining. In the last, the thesis proposes a new clustering algorithm based on grid. The experimental results have proved the feasibility and better performance of this algorithm.The non-linear relationship between the speed of similar case retrieval and the scale of case base is one of the specialties of the similar case retrieval method in Case-based reasoning System. Considering the characteristics of CABG (Clustering Algorithm Based on Grid) which is proposed in this thesis, a new case-based reasoning strategy which divides the process into two is given. On the basis of the computation of cases similarity, optimized methods of similarity computation and how to fix the center of classes are proposed. In the last, the frame of system is given and put into realization, which is also tested with the data obtained from international database. The result of experiment shows that the system has good performance... |