Track:
Contents
Downloads:
Abstract:
In a variety of reasoning tasks, even ones for which CBR seems ideally suited, a stand-alone CBR component may not prove adequate. First, the data available in system construction may be too raw or noisy for direct processing and may require sophisticated reasoning before it is in a form suitable for CBR. Second, capacity demands and other run-time constraints may prohibit a straight CBR module from being deployed. This paper describes a pipelined architecture where one or more reasoning steps are used to preprocess data into a form suitable for use in CBR, and CBR is used as a synthesis component for the creation of a stand-alone, run-time database.