An extraction method of analogical information for efficient knowledge processing

Case-based reasoning is to solve new problems by adapting previous solutions to old problems. As an alternative to rule-based reasoning, the extraction of similar examples from previous knowledge is very important for solving practical problems. In this paper, we propose a method to extract analogical information for efficient knowledge processing in an intelligent information system supported by relational database technology. At first, the method incrementally retrieves candidate analogical cases from the case base. Then, it selects more similar cases, based on the similarity between the problem and the candidate. The approach substantially reduces the computational complexity in knowledge processing of intelligent information systems such as expert systems.<>