Kevin D. Ashley and Vincent Aleven
Before people can decide whether to rely on the advice provided by a Case-Based Reasoning (CBR) system, they must understand the criteria according to which the system asserts that a case is relevant to a problem and that it is more relevant than other cases. In our application, tutoring students to reason with cases, our intelligent tutoring system, CATO, needs to explain its relevance criteria and illustrate them with examples. Moreover, it needs to deal with a variety of relevance criteria, some of which involve relations among multiple cases. The system designers, and eventually, teachers and students, need to be able to understand, use and modify the program’s concepts for assessing case relevance and constructing case-based arguments. A logical representation of the relevance criteria provides the expressiveness and flexibility to make that possible.