New Algorithms for Knowledge Automation of CBR Retrieval and Adaptation

Recently, Case-Based Reasoning (CBR) has proved its success as reasoning and learning approach. However, there are some knowledge engineering complexity appears in developing the CBR systems. This paper introduces a new CBR system that helps to reduce the knowledge acquisition effort required for building typical CBR systems. The proposed system incorporates the learning techniques into the CBR methodology to automate extracting the features weights of the cases, and to extract the adaptation rules from the case library. This improves the performance of CBR systems by eliminating the need for expert to guide these developing steps, especially for the situations where a little knowledge of the field is known. Also, it increases the accuracy of the achieved solution of the problem to be solved. The proposed system proves its performance when applying for real systems.