CAREER: General And Optimal Layered Network Visualization

该奖项的全部或部分资金来自《2021年美国救援计划法案》(公法117-2)。医学、计算机科学、物理学和社会学等领域的人类决策依赖于结构化数据。长期以来，网络一直是计算机科学中组织数据的常见模型。然而，直到最近，网络可视化才成为更常用的工具；例如，出版商开始在新闻文章中使用网络可视化。分层网络可视化显示网络随时间的变化或状态之间的流动。它们有助于可视化复杂的动力学，例如水如何通过蒸汽机的部件。创建基于人类可读性标准的最佳布局是一个开放的挑战。创建可视化的现有算法将计算速度优先于人类可读性。该项目的目标是为分层网络可视化创建通用的最佳布局算法，从而优先考虑可读性。最优布局对于可读性非常重要的领域应用尤其有益，例如，在医学领域，错误可能会产生严重后果。通过综合研究和教育，该项目将激励学生通过将应用研究和基础研究相结合来解决有意义的人类问题；吸引和培训下一代可视化科学家，重点关注未被充分代表的群体；广泛传播研究成果，以帮助未来的研究人员和设计师创建更有效的可视化数据探索和决策工具。这项研究将推进分层网络可视化的最新水平，并导致更有效的可视化数据探索和决策工具。它将为关键任务提供最佳布局，并为评估布局启发式提供基线。五项形成性研究将产生分层网络可视化的一般要求语料库，这将指导未来的研究人员和实践者。这些要求将被集成到用户可定制的最佳布局算法中。该算法以及近似和启发式布局方法将在用于分层网络可视化的免费开放源码库中发布。特定领域和一般的计算和人类受试者研究将提供证据支持的设计指南，这些指南可以跨域转移。实验将从人类可读性和技术可行性两个方面定义可伸缩性限制。该项目将通过提供基准数据集、评价方法和最佳基线实施，帮助评估未来的布局算法。应用研究和基础研究相结合的研究将引导职业生涯早期的研究生在跨学科团队中研究亲社会问题的出版物和职业。该项目将通过夏令营活动吸引科学、技术、工程和数学领域代表性不足的6-12年级学生参加，并开发教学材料，用于在没有预先可视化组件的课程中教授研究生。研究成果将在学术场所、在线和通过领域合作者公开传播。为了鼓励快速和不受限制的采用以及可复制、可复制的研究，所有产品-如预印本、源代码、研究材料和数据集-将以可靠的长期档案免费和开源发布。该奖项反映了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 award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).Human decision-making in fields such as medicine, computer science, physics, and sociology rely on structured data. Networks have long been common models in computer science for structuring data. However, only recently network visualization has become more commonly used tool; for instance, publishers are starting to use visualizations of networks in news articles. Layered network visualizations show either changes in networks over time or flows between states. They help visualize complex dynamics such as how water moves through the parts of a steam engine. Creating an optimal layout based on human readability criteria is an open challenge. Existing algorithms creating visualizations prioritize computational speed over human readability. The goal of this project is to create general and optimal layout algorithms for layered network visualizations that prioritize readability instead. Optimal layouts are particularly beneficial for domain applications where readability is paramount, e.g., in medicine where mistakes can have significant consequences. Through integrated research and education, the project will motivate students to work on meaningful human problems by combining applied and basic research; attract and train the next generation of visualization scientists, with a focus on people from underrepresented groups; and widely disseminate research results to help future researchers and designers create more effective visual data exploration and decision-making tools.This research will advance the state-of-the art in visualizing layered networks and lead to more effective visual data exploration and decision-making tools. It will provide optimal layouts for critical tasks and baselines for evaluating layout heuristics. Five formative studies will produce a corpus of general requirements for layered network visualization which will guide future researchers and practitioners. These requirements will be integrated into a user-customizable optimal layout algorithm. The algorithm, as well as approximate and heuristic layout approaches, will be released in a free and open-source library for layered network visualization. Domain-specific and general computational and human-subjects research will provide evidence-backed design guidelines that are transferable across domains. Experiments will define scalability limitations both in terms of human readability and technical feasibility. This project will contribute to assessing future layout algorithms by providing benchmark datasets, evaluation methodologies, and optimal baseline implementations. Combined applied and basic research studies will lead early-career graduate students to publications and careers working on prosocial problems in cross-disciplinary teams. The project will engage grade 6--12 students from groups underrepresented in science, technology, engineering, and mathematics through summer camp activities and develop instructional materials for teaching graduate students in courses without prior visualization components. Research outcomes will be publicly disseminated at academic venues, online, and through domain collaborators. To encourage rapid and unrestricted adoption and replicable, reproducible research all products---such as preprints, source code, study materials, and datasets---will be released free and open-source on a reliable long-term archive.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.