DDRIG: Language of scientific uncertainty and risk in food safety and environmental science

这个项目着眼于政府关于食品安全和环境风险的报告，以比较对风险和不确定性的讨论。所研究的报告将涵盖对公众健康的潜在危害。利用二十年的科学政策报告，该项目将分析科学传播中不确定性的不同表现。在谈到风险和不确定性时，对监管文件的分析将有助于保护公众。该项目将引起科学家、政策制定者和公众的兴趣。该项目将汇编一个包含监管机构发表的科学报告中不确定性表达的句子的数据集。它将对联邦机构委托进行的特定危险风险评估中的不确定性表达进行比较分析。研究人员将开发一种机器学习自然语言处理(NLP)模型，以帮助对过去20年出版的文件进行人工监督编码。编码将侧重于通过特定形式的不确定性的分类来表达。该项目将产生纵向案例，以供在两个政策背景下进行比较。该项目的定量比较分析部分将说明知识形成和不确定性表达在政府食品安全科学的认知文化中的相互作用。该项目将使用NLP系统的能力来检测环境风险评估中围绕危害表达的不确定性。该项目的发现将为使用新兴机器学习技术进行社会科学内容分析的适宜性和风险的讨论提供信息。该项目将产生一个人类验证的数据集，其中包括科学家对影响公众健康的危险的不确定性。通过使用公开发布的政府风险评估，由此产生的不确定性表达数据库也可以在网上发布，并向广大公众提供。这将允许感兴趣的或关心的公民检查专家关于他们的健康的各种风险的问题是如何随着时间的推移而变化的。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This project looks at government reports of food safety and environmental risk to compare discussions of risk and uncertainty. The reports studied will cover potential hazards to the public health. Using two decades of scientific policy reports the project will analyze different expressions of uncertainty within science communications. The analysis of regulatory documents will help protect the public when speaking of risks and uncertainty. The project will be of interest to scientists, policy makers, and the public. This project will assemble a dataset of sentences containing expressions of uncertainty within scientific reports published by regulatory agencies. It will do a comparative analysis of uncertainty expression within hazard-specific risk assessments commissioned by federal agencies. The researchers will develop a machine learning natural language processing (NLP) model to assist the human-supervised coding of documents published in the last two decades. The coding will focus on expressions by a taxonomy of specific forms of uncertainty. This project will yield longitudinal cases for comparison across two policy contexts. The quantitative comparative analysis component of the project will illustrate the interaction of knowledge formation and uncertainty expression within the epistemic cultures of governmental food safety science. The project will use the NLP system’s capabilities to detect uncertainties expressed around hazards in environmental risk assessment. The project’s findings will inform discussions of the suitability and risks of using emerging machine learning technologies for content analysis in the social sciences. This project will produce a human-verified dataset of scientists’ uncertainties around hazards that bear on the public’s health. By using governmental risk assessments which are openly published, the resulting database of uncertainty expressions can also be published online and made available to the public at large. This will allow interested or concerned citizen to examine how experts’ questions around the various risks to their health, stemming from their food or their environment, have changed over time.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.