| Tutorial dialogue systems interact with students using rich natural language, in support of a learning task. These systems have been shown to be effective at supporting student learning, in some cases rivaling the effectiveness of human tutors. However, due to the engineering effort required within their various components, building and developing tutorial dialogue systems is a challenging task. To address this challenge, machine learning approaches that automatically learn from human-human dialogue are the focus of increasing attention. One of the most important tasks for a tutorial dialogue system is to understand students' natural language input in order to respond accordingly. This dissertation research focuses on techniques for understanding dialogue acts, which are used to represent the intention of an utterance. |
