Maintaining retrieval knowledge in a case-based reasoning system.

摘要: The knowledge stored in a case-base is central to the problem-solving of Case-Based Reasoning (CBR) system. Therefore, main- tenance key component maintaining CBR However, other sources, such as indexing and similarity for im- proved case retrieval, also play an important role problem-solving. For many applications refinement this retrieval necessary maintenance. This paper focuses on optimisation parameters feature selections/weights nearest-neighbour algorithms used by retrieval. Op- timisation applied after maintenance refines reflect changes that have occurred cases case-base. process generic automatic, using con- tained cases. In we demonstrate its effectiveness real tablet formulation application two scenarios. One scenario, growing case-base, provided snap-shots formula- tion database. A change companys policy results second, more fundamental, requirement We show that, maintenance, system did indeed benefit from refining knowledge. believe existing shells would including option automatically optimise process.