SaTC: CORE: Small: Effective Design and Recommendation for Privacy-Preserving Data Visualizations

该项目将开发新的技术和软件工具，以支持可视化的创建，从而保护底层数据集中敏感信息的隐私。在当今社会，可视化是一种重要的信息交流技术。不幸的是，可视化可能与数据隐私的概念不一致，因为数据集中的项目可能包含敏感信息，因此无法完全公开。即使没有发布底层原始数据，也必须仔细设计面向公众的可视化，使其不允许披露或推断数据记录中的敏感信息。这个项目的目标是使人们能够轻松地从包含敏感数据的数据集中创建设计良好、面向公众、保护隐私的可视化。项目团队将开发优化可视化设计的机制和软件（基于既定的理论和原则），同时考虑到数据中可能存在的隐私风险。新的技术和工具将作为独立的开源项目和/或集成到流行的工具包和库中供公众使用。为了实现这一目标，项目团队将开发一种可视化语法，一种用于指定可视化设计的机制，而不需要低级编程或数据预处理，它支持隐私约束和清理操作。该语法将用于在隐私风险背景下建立有效可视化设计的理论和指导方针，通过进行大规模的众包实验来评估不同的设计选择、隐私规则和清理技术行为如何影响最终可视化的效用和隐私。开发的理论和指导方针的影响将被转化为工具，这些工具既可以推荐保护隐私的可视化，也可以支持创作和改进具有隐私意识的可视化。此外，该团队将开发大学和高中水平的教育材料，利用项目中创建的发现和工具来提高可视化中的隐私意识。该奖项反映了国家科学基金会的法定使命，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。

This project will develop novel techniques and software tools to support the creation of visualizations that preserve the privacy of sensitive information in the underlying dataset. Visualization is an important technique for communicating information in our society. Unfortunately, visualization can stand at odds with the concept of data privacy, where items in a dataset cannot be fully made public due to the existence of sensitive information that they might contain. Even when the underlying raw data is not published, public-facing visualizations must be carefully designed so that they do not allow the disclosure or inference of sensitive information in the data records. The goal of this project is to make it easy for people to create well-designed, public-facing, privacy-preserving visualizations from datasets that contain sensitive data. The project team will develop mechanisms and software that optimize the design of visualizations (based on established theory and principles) while also considering the privacy risks that might exist in the data. New techniques and tools will be released for public use, as standalone open source projects and/or integrated into popular toolkits and libraries.To accomplish this goal, the project team will develop a visualization grammar, a mechanism for specifying visualization designs without requiring low-level programming or data pre-processing, that supports privacy constraints and sanitization actions. The grammar will be used to establish theories and guidelines for effective visualization design in the context of privacy risks, via conducting large-scale, crowdsourced experiments that assess how different design choices, privacy rules, and sanitization techniques actions affect utility and privacy of the resulting visualizations. The impact of the developed theory and guidelines will then be translated into tools that both recommend visualizations that preserve privacy and support the authoring and refining of privacy-aware visualizations. Further, the team will develop educational materials at both the college and high school level that leverage the findings and tools created in the project to raise awareness of privacy in visualization,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.