Learning Case Adaptation
Laptop models of case-based reasoning (CBR) generally guide case version using a fixed set of adaptation rules. A hard practical issue is how to recognize the knowledge necessary to guide adaptation for particular tasks. Similarly, an open issue for CBR as a cognitive model can be how circumstance adaptation understanding is discovered. We identify a new way of acquiring case adaptation know-how. In this way, adaptation trouble is initially fixed by thinking from scratch, applying abstract guidelines about structural transformations and general recollection search heuristics. Traces of the processing intended for successful rule-based adaptation are stored as cases to enable future adaptation to be created by case-based thinking. When identical adaptation danger is encountered in the foreseeable future, these version cases give task- and domain-specific guidance for the case variation process. We present the tenets in the approach concerning the relationship between memory search and case version, the memory search procedure, and the storage space and recycle of situations representing adaptation episodes. These kinds of points will be discussed inside the context of ongoing study on CALL, a computer model that learns case version knowledge pertaining to case-based catastrophe response planning.
The fundamental rule of case-based reasoning (CBR) for problem-solving is that fresh problems are resolved by finding stored information of before problem-solving symptoms and changing their approaches to fit new situations. Generally in most case-based thinking systems, the case adaptation method is guided by set case version rules. Working experience developing CBR systems has demonstrated that it is difficult to establish appropriate case adaptation rules (e. g., Allemang, 1993; Leake, 1994). In defining variation rules, the problem is typical operationality/generality tradeoff that was initially observed in exploration on explanation-based learning (e. g., Segre, 1987): Specific rules are super easy to apply and are also reliable, although only apply to a narrow range of edition problems; abstract rules course a broad array of potential modifications but are often hard and expensive to utilize because they just do not provide task- and domain-specific guidance. In those CBR systems which in turn perform circumstance adaptation, certain rules are usually used, demanding that the creator perform challenging analysis from the task and domain to determine which guidelines will be needed. In practice, the difficulties of determining adaptation guidelines are so acute that many CBR applications simply omit case adaptation (e. g., Barletta, 1994).
This paper reveals a new method by which a case-based reasoning program can study adaptation understanding from experience.