Learning From and Reasoning About Case-based Reasoning Systems

This paper gives a survey about existing case-based reasoning systems and what can be learned from their development and usage. The information gathering for this survey has been carried out by sending two structured questionnaires to developers of case-based reasoning systems. On the basis of 50 answers coming from all over the world we classify and summarise the results concerning the cases acquired and used, the tasks fulfilled, the background knowledge needed and interesting experiences from bringing the systems into use. The information gathered enables us to identify those steps in the CBR cycle [AaPl94] that require further research and development. Finally we give an outlook to the next step currently underway that is directed towards making the collected CBR cases available in a CBR system that is accessible via WWW. By means of this 'meta CBR system' we give the CBR community an opportunity to learn from experience through using the CBR technology for the improvement of our o wn field.