RI-Medium: Collaborative Research : Corpus-based Studies of Lexical, Acoustic, And Discourse Entrainment in Spoken Dialogue

RI-Medium：协作研究：基于语料库的口语对话中的词汇、声学和话语夹带研究

• 批准号: 0803159
• 负责人: Ani Nenkova
• 金额: $ 34.38万
• 依托单位: University of Pennsylvania
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2008
• 资助国家: 美国
• 起止时间: 2008-09-01 至 2013-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0803159&HistoricalAwards=false
• 关键词: RI; Medium; Collaborative; Research; Corpus

Participants in human-human conversation often entrain to one another, adopting thevocabulary and other behaviors of their partners.Evidence of this has been found from laboratory studies and observations of real lifesituations. We are investigating many types of entrainment in two large corpora ofhuman-human conversations to improve system behavior in Spoken Dialogue Systems(SDS). We want to discover which types of entrainment occur generally acrossspeakers and which seem to be speaker-specific, which types of entrainment can bereliably linked to task success and perceived naturalness, and which types ofentrainment can be automatically modeled in SDS.Our research has importance for the construction of better SDS.Currently, research SDS have attempted to entrain users to system vocabularies toimprove speech recognition accuracy: Since users are likely to employ the samevocabulary in their answers that systems use in their queries, systems have a betterchance of recognizing user input correctly if they can predict word usage. However,there has been little attempt to create SDS that entrain to user behavior, despiteevidence that human beings rate humans and systems that behave more like themmore highly than those that do not. Our work focuses on determining which types ofsystem entrainment to users will be most important to users and most feasible for SDS.Our results will be disseminated through papers and presentations at speech andlanguage conferences. We will also provide publicly available annotated corpora forfuture research by others.

人与人之间对话的参与者经常相互吸引，采用对方的词汇和其他行为方式。实验室研究和对真实的生活环境的观察都证明了这一点。我们正在研究在两个大型语料库的人与人的对话，以改善系统的行为口语对话系统（SDS）的夹带的多种类型。我们希望发现哪些类型的夹带通常发生在说话者之间，哪些类型的夹带似乎是特定于说话者的，哪些类型的夹带可以可靠地与任务成功和感知自然度联系起来，以及哪些类型的夹带可以在SDS中自动建模。我们的研究对于构建更好的SDS具有重要意义。目前，研究SDS试图将用户引入系统词汇表以提高语音识别准确率：由于用户可能会在他们的答案中使用与系统在查询中使用的相同的词汇，如果系统可以预测单词的使用，则系统正确识别用户输入的机会更大。然而，几乎没有人试图创建一个能影响用户行为的SDS，尽管有证据表明，人类对行为更像他们的人和系统的评价要高于那些不像他们的人和系统。我们的工作重点是确定哪些类型的系统夹带到用户将是最重要的用户和最可行的SDS。我们的结果将通过论文和演讲和语言会议上的演示文稿传播。我们还将提供公开的注释语料库，供其他人将来研究。