Building a CBR system still remains a difficult task due to the difficulties of developing suitable retrieval and adaptation mechanisms for a given application. To address these difficulties, we extended the basic Ripple Down Rules framework to allow the incremental development of an adaptation function during the use of the system for solving actual problems. In our approach the expert is only required to provide explanations of why, for a given problem, a certain adaptation step should be taken. Incrementally a complex adaptation function as a systematic composition of many simple adaptation functions is developed. Our approach is effective with respect to both, the development of highly tailored and complex adaptation functions for CBR as well as the provision of an intuitive and feasible approach for the expert. The approach has been implemented in our CBR system MIKAS, for the design of menus according to dietary requirements. In this paper we present experimental evidence for the suitability of our approach to address the adaptation problem in the development of CBR systems.
展开▼