III: Medium: Ask the Experts: Generating Question-Answer Pairs for Addressing Information Deficits about Vaccines

III：媒介：询问专家：生成问答对以解决疫苗信息缺陷

• 批准号: 2211526
• 负责人: Bhuwan Dhingra
• 金额: $ 114.42万
• 依托单位: Duke University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-09-01 至 2026-08-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2211526&HistoricalAwards=false
• 关键词: III; Medium; Ask; Experts; Generating

This project aims to support decision-making by developing a scalable pipeline for identifying information deficits. In this context, deficits refer to topics where there is a high demand for credible information but the corresponding supply is low. The pipeline begins with developing automated methods for identifying emerging and extant vaccine-related concerns from online sources. The moves to the creation of generalized questions that exemplify the concerns and then match the questions to credible answers. The techniques developed as part of this research will facilitate decision-making in areas with an ever-evolving knowledge base of contextualized answers to challenging questions. The research will focus on addressing information deficits about vaccines as the target application domain. The pandemic and the rapid development of vaccines has resulted in many questions from the public. For example, has the COVID-19 vaccine led to more deaths than all other vaccines put together? As questions about vaccines continue to propagate in online communication channels, timely and accurate answers to them are key to supporting decision-making.The main thrusts of this research will be to develop methods for: (1) identifying emerging and extant vaccine-related concerns from online sources, such as social media and blogs, and categorizing them using a novel taxonomy; (2) generating contextualized questions from text aimed at addressing specific claims, concerns, knowledge gaps, and myths; and (3) matching the questions to credible answers when they exist, or otherwise recommending them to experts who can develop new answers. In addition to producing broadly applicable research results and knowledge artifacts, this project, guided by an expert advisory group on vaccinations, will develop a public-facing dashboard aimed at assisting public health and medical professionals by surfacing ongoing discussions about vaccines and highlighting the areas that require attention.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该项目旨在通过开发可扩展的管道来识别信息缺陷来支持决策。在这种情况下，赤字是指对可信信息的需求较高但相应的供应量较低的主题。该管道首先开发自动化方法，用于从在线来源识别新出现和现有的疫苗相关问题。创建概括性问题来举例说明问题，然后将问题与可信答案相匹配。作为这项研究的一部分开发的技术将有助于在具有不断发展的知识库的领域中做出决策，这些知识库可以针对挑战性问题提供情境化答案。该研究将重点解决作为目标应用领域的疫苗的信息缺陷。大流行和疫苗的快速发展引起了公众的许多疑问。例如，COVID-19 疫苗导致的死亡人数是否比所有其他疫苗加起来还要多？随着有关疫苗的问题不断在在线沟通渠道中传播，及时、准确地回答这些问题是支持决策的关键。这项研究的主要目标将是开发以下方法：（1）识别新出现和现有的疫苗相关问题来自社交媒体和博客等在线资源，并使用新颖的分类法对它们进行分类； (2) 从文本中生成情境化问题，旨在解决具体主张、关切、知识差距和神话； (3) 将问题与存在的可信答案进行匹配，或者以其他方式将其推荐给可以得出新答案的专家。除了产生广泛适用的研究成果和知识工件外，该项目在疫苗接种专家咨询小组的指导下，还将开发一个面向公众的仪表板，旨在通过呈现有关疫苗的持续讨论并强调需要关注的领域，为公共卫生和医疗专业人员提供帮助。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Interface Design for Crowdsourcing Hierarchical Multi-Label Text Annotations

• DOI: 10.1145/3544548.3581431
• 发表时间: 2023-02
• 期刊: Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems
• 作者: Rickard Stureborg;Bhuwan Dhingra;Jun Yang