SCC-PG: Privacy and Fairness in Planning when using Third-Party, Heterogeneous Data Sources

今天的城市和社区规划能够从居民主动和被动参与数据收集所产生的大量移动和位置数据中汲取经验。移动性信息可以从许多来源收集，这些来源通常由第三方收集和维护，以用于多个异构应用程序。然而，当结合起来时，这些第三方的异构数据可以描绘出城市景观中丰富的运动画面，这可能是社区规划者的重要信息来源。然而，通过机器学习等方式将异构数据源天真地纳入其中，带来了许多技术和社会挑战，这些挑战影响了城市居民的隐私，并引发了有关公平和社会正义的重要问题。这些数据在社区中的分布并不均匀，可能会遗漏重要的人群并使算法决策产生偏差，而且非常详细的位置信息可能会泄露有关应该受到保护的居民的信息。 作为该规划补助金的一部分，该项目将解决与隐私相关的基础研究挑战，以及考虑到居民参与和规划决策的社区规划异构数据应用的公平性。包括：（1）我们如何识别和减轻社区规划中可能来自异构的第三方数据源的潜在偏见？(2)我们如何设计隐私感知算法，可以合并不同和不完整的数据，同时为社区规划提供有意义的结果？(3)我们如何将公平和偏见的联合收割机原则与隐私保护数据分析方法结合起来？(4)我们如何识别数据源中的代表性不足，并探索有助于参与的因素？(5)居民和社区规划者如何理解移动数据的隐私？此外，该项目将寻求扩大社区合作伙伴关系和外联在规划赠款过程中，通过社区讲习班，以确定基础研究的关键方面，以及如何将成果融入实践。该奖项反映了NSF的法定使命，并已被认为是值得支持的评估使用基金会的知识价值和更广泛的影响审查标准。

Civic and community planning today is able to draw from a wealth of mobility and location data generated by residents who both actively and passively participate in data collection. Mobility information can be gleaned from a number of sources, which are often collected and maintained by third-parties for uses in multiple, heterogeneous applications. However, when combined, these third-party, heterogeneous data can paint a rich picture of movement through the urban landscape, that could be an important source of information for community planners. However, the naive inclusion of heterogeneous data sources, such as through machine learning, presents a number of technical and social challenges that impact the privacy of city residents and raise important questions around fairness and social justice. The data is not uniformly distributed across communities, potentially leaving out important populations and biasing algorithmic decision making, and the highly detailed location information may leak information about residents that should be protected. As part of this planning grant, this project will address basic research challenges associated with privacy as well as fairness in the application of heterogeneous data for community planning that take into account both resident participation and planner decision making. Including: (1) How do we both identify and mitigate potential biases in community planning that can arise from heterogeneous, third-party data sources? (2) How do we design privacy-aware algorithms that can merge disparate and incomplete data while providing meaningful results for community planning? (3) How do we combine principles of fairness and bias with privacy preserving data analysis methods? (4) How do we identify under-representation within data sources and explore factors that contribute to participation? (5) How do residents and community-planners understand privacy with respect to mobility data? Additionally, the project will seek to expand community partnership and outreach during the planning-grant process through community workshops to identify key aspects of the basic research and how the outcomes can be integrated into practice.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.