REU SITE: From Formal Computer Science Education to Real World Data Science Research to Policy Decision Making

REU 站点：从正规计算机科学教育到现实世界数据科学研究再到政策决策

• 批准号: 2244271
• 负责人: Lisa Singh
• 金额: $ 44.38万
• 依托单位: Georgetown University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-04-01 至 2026-03-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2244271&HistoricalAwards=false
• 关键词: REU; SITE; Formal; Computer; Science

When attempting to tackle societal scale issues, computer science students have limited opportunities to use algorithms, methods, and the tools they are taught in class with large-scale real world data sets. It is even more unusual to get the opportunity to understand how to build bridges that connect these algorithms and methods to policy development and decision making. This program aims not only to teach students about computer science research, but to also help them understand how researchers in other disciplines use computer science algorithms and analytic tools to generate evidence for developing public policy. The Georgetown REU site connects formal computer science education to real world data science research to public policy decision making. Students in the program work on improving mining and learning algorithms for different data science algorithms. They then connect the outputs of the methods they develop to social science and public policy questions, improving their understanding of how those outside of computer science use algorithms and their specifications to generate scientific evidence for developing public policy.The core research that the REU students conduct is in data-centric computing. Their goals are to advance the state-of-the-art methods for emotion detection across languages (cohort 1), emerging misinformation detection (cohort 2), and opinion modeling of public policies (cohort 3). All three of these problems have existing solutions. Each year students extend the existing state of the art methods to address one specific constraint. Cohort 1 employs multi-lingual language models to tackle the language constraint. Cohort 2 generates novel weakly labeled data to address the temporal (emerging) constraint. Cohort 3 focuses on blending auxiliary information to address the limited training data constraint. For all three tasks, students also consider unsupervised, semi-supervised, and supervised models and conduct sensitivity analyses that identify the biases associated with different learning techniques, extending their understanding of the tradeoffs between interpretability, scalability, and accuracy. Finally, students use the “best” models to conduct a data science analysis that informs public policy research with respect to migration movement, misinformation intervention strategies, and gun culture. The Georgetown REU Site gives students the opportunity to advance computer science research and understand how computer science research connects to other disciplines of research.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.

当试图解决社会规模的问题时，计算机科学专业的学生只有有限的机会使用他们在课堂上教授的算法，方法和工具，以及大规模的真实的世界数据集。更难得的是，有机会了解如何建立桥梁，将这些算法和方法与政策制定和决策联系起来。该计划的目的不仅是教学生有关计算机科学研究，但也帮助他们了解其他学科的研究人员如何使用计算机科学算法和分析工具，以生成制定公共政策的证据。乔治城REU网站将正式的计算机科学教育与真实的世界数据科学研究和公共政策决策联系起来。该计划的学生致力于改进不同数据科学算法的挖掘和学习算法。然后，他们将他们开发的方法的输出与社会科学和公共政策问题联系起来，提高他们对计算机科学之外的人如何使用算法及其规范来为制定公共政策生成科学证据的理解。REU学生进行的核心研究是以数据为中心的计算。他们的目标是推进跨语言情感检测（队列1），新兴错误信息检测（队列2）和公共政策意见建模（队列3）的最先进方法。所有这三个问题都有现有的解决方案。每年学生扩展现有的最先进的方法，以解决一个特定的限制。队列1采用多语言语言模型来解决语言约束。队列2生成新的弱标记数据，以解决时间（新兴）约束。队列3侧重于混合辅助信息，以解决有限的训练数据约束。对于所有这三个任务，学生还考虑无监督，半监督和监督模型，并进行敏感性分析，识别与不同学习技术相关的偏差，扩展他们对可解释性，可扩展性和准确性之间权衡的理解。最后，学生使用“最佳”模型进行数据科学分析，为移民运动，错误信息干预策略和枪支文化方面的公共政策研究提供信息。乔治城REU网站为学生提供了推进计算机科学研究的机会，并了解计算机科学研究如何与其他学科的研究相联系。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。