| Abstract
|
In this paper I describe an approach to au … In this paper I describe an approach to automated knowledge acquisition inwhich users specify desired system behavior by constructing justifications ofexamples. Justifications are explanations of why example behaviors areappropriate in given situations. I analyze the problem of acquiringjustifications, showing how current knowledge acquisition techniques are bestsuited for asking what-questions, while justifications are naturally viewed asanswers to why-questions. I sketch a new approach for acquiringjustifications that transforms why-questions into what-questions, borrowingthe sources of power of existing techniques. In this approach, usersconstruct justifications by selecting facts that specify what is relevant in asituation from a space of facts provided by the elicitation tool.Justifications are then used to create operational mappings from situations tointended outcomes. I show how the approach is applied to two differentknowledge acquisition problems: the acquisition of diagnostic strategy andthe acquisition of design rationale. I conclude by identifying commoncharacteristics of the two applications and discuss how their designdistributes the cognitive load between human and machines. cognitive load between human and machines.
|
