DISCOURSE GENERATION FOR INSTRUCTIONAL APPLICATIONS - IDENTIFYING ANDEXPLOITING RELEVANT PRIOR EXPLANATIONS

To reap the benefits of natural language interaction, tutorial systems must be endowed with the properties that make human natural-language interaction so effective. One striking feature of naturally occurring interactions is that human tutors and students freely refer to the con text created by prior explanations. In contrast, computer-generated ut terances that do not draw on the previous discourse often seem awkward and unnatural and may even be incoherent. The explanations produced b y such systems are frustrating to students because they repeat the sam e information over and over again. Perhaps more critical is that, by n ot referring to prior explanations, computer-based tutors are not poin ting out similarities between problem-solving situations and therefore may be missing out on opportunities to help students form generalizat ions. In this article, we discuss several observations from an analysi s of human-human tutorial interactions and provide examples of the way s in which tutors and students refer to previous explanations. We desc ribe how we have used a case-based reasoning algorithm to enable a com putational system to identify prior explanations that may be relevant to the explanation currently being generated. We then describe two com putational systems that can exploit this knowledge about relevant prio r explanations in constructing their subsequent explanations.