A study of an automatic learning model of adaptation knowledge for case base reasoning

This paper proposes an algorithm to acquire adaptation knowledge through case comparison in order to extract connoted knowledge that is contained in cases. To acquire adaptation knowledge from cases in case base, comparable cases are selected. This paper also propose a similarity method for the selection of comparable cases.For simulation, the case of deciding automobile price is considered to look into how much the acquired adaptation rules through case comparison are effective in improving the precision of problem solving. Simulation results show that a system using the extracted adaptation knowledge provided higher precision of solution than a system supporting only case extraction, and even when the loss value of case attributes and case solutions were inaccurate, the acquired adaptation rules could improve the precision.