tailieunhanh - Báo cáo khoa học: "Learning to Interpret Utterances Using Dialogue History"

We describe a methodology for learning a disambiguation model for deep pragmatic interpretations in the context of situated task-oriented dialogue. The system accumulates training examples for ambiguity resolution by tracking the fates of alternative interpretations across dialogue, including subsequent clarificatory episodes initiated by the system itself. We illustrate with a case study building maximum entropy models over abductive interpretations in a referential communication task. | Learning to Interpret Utterances Using Dialogue History David DeVault Institute for Creative Technologies University of Southern California Marina del Rey CA 90292 devault@ Matthew Stone Department of Computer Science Rutgers University Piscataway NJ 08845-8019 Abstract We describe a methodology for learning a disambiguation model for deep pragmatic interpretations in the context of situated task-oriented dialogue. The system accumulates training examples for ambiguity resolution by tracking the fates of alternative interpretations across dialogue including subsequent clarificatory episodes initiated by the system itself. We illustrate with a case study building maximum entropy models over abductive interpretations in a referential communication task. The resulting model correctly resolves 81 of ambiguities left unresolved by an initial handcrafted baseline. A key innovation is that our method draws exclusively on a system s own skills and experience and requires no human annotation. 1 Introduction In dialogue the basic problem of interpretation is to identify the contribution a speaker is making to the conversation. There is much to recognize the domain objects and properties the speaker is referring to the kind of action that the speaker is performing the presuppositions and implicatures that relate that action to the ongoing task. Nevertheless since the seminal work of Hobbs et al. 1993 it has been possible to conceptualize pragmatic interpretation as a unified reasoning process that selects a representation of the speaker s contribution that is most preferred according to a background model of how speakers tend to behave. In principle the problem of pragmatic interpretation is qualitatively no different from the many problems that have been tackled successfully by data-driven models in NLP. However while researchers have shown that it is sometimes possible to annotate corpora that capture features of in terpretation to .