CRII: CHS: Visualizing Data Relationships Across Multiple Views

在一个数据日益丰富的世界中，有效设计用于分析和可视化从卫生到国家安全等领域数据的工具具有巨大的社会和科学价值。分析师的常见情况是需要同时管理数据集的多个视图，这就提出了可视化分析工具如何帮助分析师看到不同视图之间的关系的问题。当前的工具需要大量的工作，例如在图中重复选择单个节点，手动跟踪直方图、散点图和地图中突出显示的信息，以识别异常行为。该项目将研究分析工具的方法，以灵活而不显眼地显示跨多个可视化的关系，以支持数据分析。这样做将加深我们对多重可视化如何支持数据语义的科学理解，并扩展多视图可视化的设计空间。这项工作还将丰富数据科学和以人为中心的计算课程的教材。该项目包括四项主要工作。首先，团队将设计一个可用的跨视图数据关系概览，由计算支持，以帮助用户从多个可视化中关联信息。这个设计突出了“上下文分离”的核心概念，它创建独立的视图来绘制计算的、跨视图的数据关系，而不是依赖于显示的视觉元素。其次，研究人员将通过考虑数据之间的关系，创建一种新的空间组织多个可视化的布局方法。第三，团队将开发新的用户交互来操纵显示的跨视图数据关系，包括合并和分割，以支持分析和导航。第四，团队将进行用户实验，研究这些跨视图数据关系可视化的有效性和认知负荷。此外，在这个项目中开发的技术将是开源的，因此研究人员、数据科学家、从业者和教育工作者可以使用、扩展或修改它们来进行数据分析、工具开发和教学。该奖项反映了美国国家科学基金会的法定使命，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。

In an increasingly data-rich world, effective design of tools for analyzing and visualizing data in domains from health to national security can have great social and scientific value. A common situation for analysts is needing to manage multiple views of a dataset at the same time, raising questions of how visual analysis tools can help analysts see relationships between separate views. Current tools require extensive effort, such as repetitively selecting individual nodes in a graph and manually following information highlighted in a histogram, scatterplot, and map to identify abnormal behaviors. This project will investigate methods for analytics tools to flexibly and unobtrusively display relationships across multiple visualizations for supporting data analytics. Doing this will deepen our scientific understanding of how multiple visualizations support sensemaking of data and expand the design space of multi-view visualizations. The work will also enrich educational materials in data science and human-centered computing courses.This project includes four major lines of work. First, the team will design a usable overview of cross-view data relationships, supported by computations, for helping users to relate information from multiple visualizations. This design highlights a core concept of "context separation", which creates stand-alone views to plot computed, cross-view data relationships instead of relying on displayed visual elements. Second, the researchers will create a novel layout method for spatially organizing multiple visualizations by considering data relationships among them. Third, the team will develop novel user interactions for manipulating displayed cross-view data relationships, including merging and splitting, to support both analysis and navigation. Fourth, the team will conduct user experiments to study the effectiveness and cognitive load of these cross-view data relationship visualizations. In addition, techniques developed in this project will be open-sourced, so researchers, data scientists, practitioners, and educators can use, extend or modify them for data analysis, tool development, and teaching.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.