Challenges in Interpreting Spoken Military Commands and Tutoring Session Responses
نویسندگان
چکیده
Various challenges have emerged over several years of grammar engineering for the spoken dialogue interface to the Navy damage control simulator DC-Train and the Spoken Conversational Tutor SCoT-DC, which reviews DC-Train performance. The systems use two methods for finding interpretations for student utterances from the recognized string. First, a Gemini grammar interprets full strings into a complex, structured logical form. A successful Gemini logical form is the preferred interpretation. Next, a robust Nuance natural language grammar looks for any interpretable phrases in the utterances which Gemini could not interpret, and uses heuristics to determine the best set of slots and values. Voice-Enabled DC-Train and SCoT-DC face challenges due to speech recognition errors, disfluent speech, underspecified responses requiring dialogue context for disambiguation, students' varying levels of familiarity and skill at using the preferred military terminology, and the need for coordination with the dialogue manager's strategies for clarification, confirmation and modeling of student uncertainty.
منابع مشابه
Discourse and Dialogue Processing in Spoken Intelligent Tutoring Systems
In recent years, the development of intelligent tutoring dialogue systems has become more prevalent, in an attempt to close the performance gap between human and computer tutors. Tutoring applications differ in many ways, however, from the types of applications for which spoken dialogue systems are typically developed. This talk will illustrate some of the opportunities and challenges in this a...
متن کاملSession 1: Spoken Language Systems
Without the ability to interpret natural language, speech recognition is suited only for a subset of tasks (though certainly not trivial ones), such as data entry, simple commands or dictation. Similarly, without speech recognition natural language is restricted to the interpretation of written language, a stylized form of human communication. Spoken language systems thus represent an attempt t...
متن کاملMon.O2b.04 Learning When to Listen: Detecting System-Addressed Speech in Human-Human-Computer Dialog
New challenges arise for addressee detection when multiple people interact jointly with a spoken dialog system using unconstrained natural language. We study the problem of discriminating computer-directed from human-directed speech in a new corpus of human-human-computer (H-H-C) dialog, using lexical and prosodic features. The prosodic features use no word, context, or speaker information. Res...
متن کاملAn I-POMDP Based Multi-Agent Architecture for Dialogue Tutoring
Dialogue systems have been widely considered as useful tools for education. The challenging tasks in developing a dialogue tutoring system include correctly interpreting student input and choosing appropriate responses. In this paper, we present a two-agent architecture for addressing the challenges. The two agents are learner agents in a reinforcement learning algorithm, which is based on the ...
متن کاملMining Free-form Spoken Responses to Tutor Prompts
How can an automated tutor assess children’s spoken responses despite imperfect speech recognition? We address this challenge in the context of tutoring children in explicit strategies for reading comprehension. We report initial progress on collecting, annotating, and mining their spoken responses. Collection and annotation yield authentic but sparse data, which we use to synthesize additional...
متن کامل