Collaborative Research: Framework for Integrative Data Equity Systems

协作研究：综合数据公平系统框架

• 批准号: 1934464
• 负责人: Julia Stoyanovich
• 金额: $ 55万
• 依托单位: New York University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-09-01 至 2022-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1934464&HistoricalAwards=false
• 关键词: Collaborative; Research; Framework; Integrative; Data

Data Science continues to have a transformative impact on Science and Engineering, and on society at large, by enabling evidence-based decision making, reducing costs and errors, and improving objectivity. The techniques and technologies of data science also have enormous potential for harm if they reinforce inequity or leak private information. As a result, sensitive datasets in the public and private sector are restricted from research use, slowing progress in those areas that have the most to gain: human services in the public sector. Furthermore, the misuse of data science techniques and technologies will disproportionately harm underrepresented groups across race, gender, physical ability, sexual orientation, education, and more. These data equity issues are pervasive, and represent an existential risk for the use of data-driven methods in science and engineering. This project will establish a Framework for Integrative Data Equity Systems (FIDES): an Institute for the study of systems that enable research on sensitive data while preventing misuse and misinterpretation. FIDES will enable interdisciplinary community convergence around data equity systems, with an initial study in critical domains such as mobility, housing, education, economic indicators, and government transparency, leading to the development of a novel data analytics infrastructure that supports responsibility in integrative data science. Towards this goal, the project will address several technically challenging problems: (1) To be able to use data from multiple sources, risks related to privacy, bias, and the potential for misuse must be addressed. This project will develop principled methods for dataset processing to overcome these concerns. (2) Individual datasets are difficult to integrate for use in advanced multi-layer network models. This project considers methods to create pre-trained tensors over large collections of spatially and temporally coherent datasets, making them easier to incorporate while controlling for fairness and equity. (3) Any dataset or model must be equipped with sufficient information to determine fitness for use, communicate limitations, and describe underlying assumptions. This project will develop tools and techniques to produce "nutritional labels" for data and models, formalizing and standardizing ad hoc metadata approaches to provenance, specialized for equity issues. In addition to supporting methodological innovation in data science, the Institute will become a focal point for sharing expertise in data equity systems. It will do so by establishing interfaces for interaction between data science and domain experts to promote expertise development and sharing of best practices, and by consistently supporting efforts on diversity and equity.This project is part of the National Science Foundation's Harnessing the Data Revolution Big Idea activity. The effort is jointly funded by the Office of Advanced Cyberinfrastructure.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.

数据科学通过实现基于证据的决策、减少成本和错误以及提高客观性，继续对科学和工程以及整个社会产生变革性的影响。如果数据科学的技术和技术加剧了不公平或泄露了私人信息，那么它们也有巨大的危害潜力。因此，公共和私营部门的敏感数据集被限制在研究用途上，减缓了那些受益最大的领域的进展：公共部门的人类服务。此外，滥用数据科学技术和技术将不成比例地损害种族、性别、体能、性取向、教育等方面未被充分代表的群体。这些数据公平问题无处不在，对于在科学和工程中使用数据驱动的方法来说，这是一个存在的风险。该项目将建立一个综合数据公平系统框架(FIDS)：一个研究系统的研究所，能够在防止误用和误解的同时研究敏感数据。FIDS将促进围绕数据公平系统的跨学科社区融合，在流动性、住房、教育、经济指标和政府透明度等关键领域进行初步研究，从而开发出支持综合数据科学责任的新型数据分析基础设施。为了实现这一目标，该项目将解决几个具有技术挑战性的问题：(1)为了能够使用来自多个来源的数据，必须解决与隐私、偏见和滥用的可能性有关的风险。该项目将开发数据集处理的原则性方法，以克服这些担忧。(2)单个数据集难以集成到高级多层网络模型中使用。这个项目考虑了在大量的空间和时间上连贯的数据集上创建预训练张量的方法，使它们更容易合并，同时控制公平和公平。(3)任何数据集或模型都必须配备足够的信息，以确定是否适合使用、传达限制并描述潜在的假设。该项目将开发工具和技术，为数据和模型制作“营养标签”，使专门针对股权问题的特别来源元数据办法正规化和标准化。除了支持数据科学的方法创新外，研究所还将成为分享数据公平制度方面专门知识的协调中心。它将通过建立数据科学和领域专家之间的互动接口来促进专业知识的发展和最佳实践的共享，并通过持续支持多样性和公平性的努力来做到这一点。该项目是国家科学基金会利用数据革命大想法活动的一部分。这项工作由高级网络基础设施办公室共同资助。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。

项目成果

Causal Intersectionality and Fair Ranking

因果交叉性和公平排名

• 发表时间: 2021
• 期刊: 2nd Symposium on Foundations of Responsible Computing (FORC
• 作者: Yang, Ke;Loftus, Joshua R.;Stoyanovich, Julia
• 通讯作者: Stoyanovich, Julia

Most Expected Winner: An Interpretation of Winners over Uncertain Voter Preferences

最受期待的获胜者：对不确定选民偏好的获胜者的解读

• 发表时间: 2023
• 期刊: Proceedings of the '23 International Conference on the Management of Data (ACM SIGMOD 2023
• 作者: Haoyue Ping;Julia Stoyanovich
• 通讯作者: Julia Stoyanovich

Taming Technical Bias in Machine Learning Pipelines

克服机器学习管道中的技术偏见

• 发表时间: 2020
• 期刊: Bulletin of the Technical Committee on Data Engineering
• 作者: Schelter, Sebastian;Stoyanovich, Julia
• 通讯作者: Stoyanovich, Julia

Data distribution debugging in machine learning pipelines

• DOI: 10.1007/s00778-021-00726-w
• 发表时间: 2022-01
• 期刊: The VLDB Journal
• 作者: Stefan Grafberger;Paul Groth;Julia Stoyanovich;Sebastian Schelter

Counterfactuals for the Future

未来的反事实

• 发表时间: 2023
• 期刊: Proceedings of the AAAI Conference on Artificial Intelligence
• 作者: Bynum, Lucius;Loftus, Joshua;Stoyanovich, Julia
• 通讯作者: Stoyanovich, Julia