casebased

Casebased, often expressed as case-based reasoning or case-based learning, refers to approaches that solve problems by recalling and adapting past cases stored in a case base. Rather than applying general rules, casebased methods rely on analogy to previously solved situations to guide decisions and actions.

In AI, case-based reasoning selects a similar past case using a similarity metric, reuses its solution, adapts it to the new context, and retains the result for future use. This cycle is commonly described as retrieve, reuse, revise, and retain (the 4Rs). Core elements are the case base, retrieval mechanisms, and adaptation strategies. Cases typically encode a problem description, the proposed solution, and the outcome, with annotations to justify decisions.

Applications of casebased methods are broad. In medicine, for diagnostic decision support and treatment planning; in software debugging and engineering; in legal and business contexts for precedent-based reasoning; and in education through case-based learning, where students analyze narratives to develop analytical skills. In education, casebased materials emphasize context and interpretation over abstract rules.

History and scope: case-based approaches emerged from cognitive science and AI research in the late 20th century as an alternative to purely rule-based systems. They have been implemented in expert systems, planning, and interactive tools, and are frequently combined with machine learning to expand and refine the case base.

Limitations include dependence on the quality and coverage of the case base, challenges in identifying truly similar cases, and the potential for ineffective or biased adaptations. Ongoing work seeks to improve retrieval, adaptation, and integration with complementary reasoning methods.