Two-way Cbr Algorithm

The task of this dissertation is to study the general CBR and the Introspective CBR and come up one of methods to resolve the problem which multi-user CBR systems face to. This method is Case-base Reasoning using two aspects learning.CBR (Case-base reasoning) is one of AI reasoning methods and based on the cases. When a new problem is came up we abstract it into a new case and find a old similar case from the case base .So we can get some useful suggestions from the old case. If the old case is not fit to the new case, then we adapt the old one to fit the new condition. If this is successful the adapted case is stored into the case base and when the same condition comes up again we will get some useful information from this stored case. This is general CBR system's principle. It only does some work in the problem domains self-learning. (E.g. in the whether forecast CBR system the problem domains self-learning is the learning of the successful whether forecast cases.) But in Case-base Reasoning using two aspects learning the problem domains self-learning is important as the system use experience self-learning (The experience users learned from the intercourse with CBR systems, such as the experience of search and adaptation.). So some new features come up and we will study, analyze and test them in this dissertation.In this dissertation, first we make some introduction of CBR system's principle and history. From this introduction we can know that although CBR system's history is not long, it has good applications in many domains. Second we specify the introspective CBR system's background, principal and features because it has recognized the importance of the system use experience learning in CBR system. Besides in this CBR system some use experience learning methods have came into being. To compare with the general CBR system we can draw a conclusion that the use experience learning methods can greatly improve the CBR system's whole efficiency. Considering this fact we came up a new CBR method, case-base reasoning using two aspects learning, in which the system use experience self-learning is important as the problem domains self-learning and the system use experience is considered as a single case base like the problem domains. This is the new method's basic thought. Third we specified and analyzed this new CBR method's principle, framework and realization, and then research the new CBR method's auto-adaptation. As a important part of CBR the auto-adaptation is researched in many CBR systems. In this dissertation we adopt the auto-adaptation based on the CBR. So there also came up some new features and the study of these new features will improve the whole CBR system's efficiency. At the same time, we adopt decision tree algorithmsto search the most similar case. The result is good.At last this dissertation give an example, the intelligent inquire system of tour, and compare the efficiency of general CBR system and the new CBR learning system. Then summarize the features of this new CBR method and draw a conclusion that this method is a feasible and efficient method to resolve the problems faced by multi-user CBR systems.