(CBR) A technique for problem
solving which looks for previous examples which are similar to
the current problem. This is useful where
heuristic
knowledge is not available.
There are many situations where experts are not happy to be
questioned about their knowledge by people who want to write
the knowledge in rules, for use in
expert systems. In most
of these situations, the natural way for an expert to describe
his or her knowledge is through examples, stories or cases
(which are all basically the same thing). Such an expert will
teach trainees about the expertise by apprenticeship, i.e. by
giving examples and by asking the trainees to remember them,
copy them and adapt them in solving new problems if they
describe situations that are similar to the new problems. CBR
aims to exploit such knowledge.
Some key research areas are efficient indexing, how to define
"similarity" between cases and how to use temporal
information.