I would link it (i.e. the CBR) with some AI systems such as expert ones.

CBR can be very knowledge intensive in that you construct cases using a lot of expert knowledge, but it can alos be very knowledge light in that you just store automatically collected data as cases. In deed, many times, CBR is considered a knowledge management process rather than a machine learning approach. Especially if you read old papers, CBR is always contrasted to expert systems rather than to machine learning. So, CBR can be considered as something positioned between completely automated approaches as machine learning and expert systems built manually.

CBR is not a restructure of these concept but has developed on its own as a solution to the knowledge acquisition bottleneck of expert systems. However, many methods from both expert systems and from machine learning/data mining can be applied when building a CBR system. In addition, it is not uncommon that the same concepts and methods have been developed in different research fields without you can claim that one is just a variant of the other. For instance, machine learning, data mining and pattern recognition are very similar but are anyway considered as different research fields.

Yes Ra'ed, CBR is considered as a new comer to AI technology. But I believe as well it can be used as part of other technologies like KM and expert systems.

Thanks Tomas, CBR yes for sure can be considered as part of many new technologies and concepts like pattern recognition, ML, KM, data mining as well.

I can see that CBR can be part of DB technology where the storage and indexing is taking place.

Can also be used as AI, ML, PR, Similarity... where in the retrieval process within CBR.

I'm wondering what is the role of CBR in mathematics?