Validation of CBL Principles

May 1999

Proceedings of the Twelfth International Florida Artificial Intelligence Research Society Conference (FLAIRS 1999)

Volume

Proceedings of the Twelfth International Florida Artificial Intelligence Research Society Conference (FLAIRS 1999)

All Papers

Case-Based Reasoning (CBR) is deemed a rather promising technology of knowledge acquisition and related knowledge processing in recent computer science. Much of the CBR’s attraction is based on the charming simplicity of its ideas. In particular, knowledge acquisition which is performed during CBR by incrementally collecting cases migth be understood as Case-Based Learning (CBL). The main goal of this paper consists in the application of validation scenarios to the investigation of principles of CBL. For this purpose, the domain of formal languages is used, where CBR is straightforwardly formalized to allow for a justified assessment of CBL principles. It is demonstrated that one of the key principles of improving case bases during CBR is not valid, seen from the viewpoint of a correct learning. Two attempts of principle refinement are undertaken and validated subsequently. For the special domain, an improvement is gained.