casebase

A casebase is a structured collection of past cases, each typically comprising a problem description and a corresponding solution. This information is used in case-based reasoning (CBR) systems, a type of artificial intelligence that solves new problems by retrieving and adapting solutions from similar past experiences. The core idea of CBR is that the most recent problems are often similar to older problems, and therefore, the solutions to older problems are likely to be relevant to newer ones.

The process of using a casebase involves several steps. First, when a new problem arises, the system searches the casebase for the most similar existing case. This similarity is determined by a defined metric that compares features of the new problem with features of stored cases. Once the most similar case is found, the system retrieves its solution. This retrieved solution is then adapted to fit the specifics of the new problem. Adaptation might involve modifying parameters, adding or removing elements, or combining solutions from multiple cases. Finally, the adapted solution is applied to solve the new problem, and if successful, this new problem-solution pair can be added to the casebase for future use, thereby growing the system's knowledge over time. Casebases are employed in various applications, including customer support, medical diagnosis, legal reasoning, and engineering design.