Case-based reasoning (CBR) solves new problems by reusing knowledge from similar past experiences. It translates human knowledge into generalized cases and applies those cases to related situations. section 2.0 section 2.6
A CBR system follows a recurring cycle:
- Retrieve relevant cases from a library by comparing important similarities and contextual factors. section 2.3
- Reuse an existing solution directly, or adapt it to fit the new problem’s differences. section 2.3
- Revise the result after solving the problem, then add the adapted case to the library for future use. section 2.3
Treating a new patient
A medical CBR system retrieves patient cases with similar symptoms, medical histories, ages, genders, and other contextual factors. It adapts an earlier treatment to the new patient, then stores the outcome—including failed outcomes—to improve later decisions and avoid repeated errors. section 2.3
CBR becomes more efficient as its case library grows, but its value depends on the relevance and quality of accumulated experience. It can also complement other techniques: fraud-prevention systems may combine historical cases with data mining, fuzzy logic, deep learning, genetic algorithms, rule-based systems, and smart agents. section 2.3 section 11.7
