Workshop: Uphill Battles in Language Technology

研讨会：语言技术的艰苦战斗

• 批准号: 1640428
• 负责人: Michael White
• 金额: $ 3.2万
• 依托单位: Ohio State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-06-15 至 2017-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1640428&HistoricalAwards=false
• 关键词: Workshop; Uphill; Battles; Language; Technology

While recent approaches to Natural Language Processing (NLP) based onmachine learning from large quantities of data have delivered valuabletechnology, this success has obscured earlier, more visionary goals ofgetting computers to understand stories, to engage in natural,cooperative dialogues with people, and to translate text and speechfluently and accurately from one human language to another. As such,this workshop highlights the value of bringing these early visionarygoals back into view. The workshop features established researchers who carriedout early work in a major sub-field of speech and language processingengaging in a moderated public dialogue with younger researchersinvolved with significant new ideas and/or methods. The workshop isorganized into four sessions, each with a moderator and invitedspeakers, on the topics of Dialogue and Speech, Natural LanguageGeneration, Text Understanding, and Grounded Language. Dialogue ineach session is enriched through questions solicited beforehand fromprospective attendees, as well as impromptu questions from theaudience. The workshop includes graduate student scribes whocollaborate to deliver a summary of each session, and a poster sessionhighlights the students' research. In addition to the workshopproceedings, the workshop's outcome is a public report based on notescontributed by the speakers and the session summaries produced by theteam of student scribes.

虽然最近基于机器从大量数据中学习的自然语言处理（NLP）方法已经提供了有价值的技术，但这种成功掩盖了让计算机理解故事，与人进行自然的合作对话，以及将文本和语音从一种人类语言流利准确地翻译成另一种人类语言的更早，更有远见的目标。因此，本次研讨会强调了重新审视这些早期远景目标的价值。研讨会的特色是在语音和语言处理的主要子领域开展早期工作的知名研究人员与涉及重要新思想和/或方法的年轻研究人员进行适度的公开对话。研讨会分为四个部分，每个部分都有一个主持人和受邀的演讲者，主题是对话和演讲、自然语言生成、文本理解和基础语言。每次会议的对话都通过事先向潜在与会者提出的问题以及观众的即兴提问而丰富。研讨会包括研究生抄写员，他们合作提供每次会议的总结，海报会议突出学生的研究。除了研讨会记录外，研讨会的成果是一份公开报告，该报告基于演讲者提供的笔记和学生抄写员团队编写的会议摘要。