Guest Editors’ Introduction Case4ased Reasoning

RE ASONING IN ARTIFICIAL INtelligence has traditionally involved writing and customizing rules or models to solve problems. But there is another paradigm that has proven effective in many experimental and applied systems: Casebased reasoning “remembers” previous problems and either adapts their solutions or uses their outcomes to evaluate new cases. There are two kinds of case-based reasoners: problem solvers and interpretive reasoners. A problem solver adapts old solutions to new problems; old solutions can provide almost-correct solutions and can warn of potential mistakes or failures. An interpretive reasoner uses cases to evaluate or justify new situations, much as lawyers use previous cases as arguments for categorizing new situations. Interpretive reasoners can evaluate solutions when no clear-cut methods are available and can interpret situations whose boundaries are open-ended or fuzzy. CASE-BASED REASONERS %EMEMBER” PREVlOUS