EAGER: Understanding and Mitigating Misinformation in Visualizations on Social Media

在危机时刻，例如在飓风或全球大流行期间，社交媒体是普通民众的重要信息来源。在这些场景中，数据可视化通常用于传达对个人决策至关重要的信息。例如，飓风路径的可视化可以告知受影响人口需要准备或疏散；而关于某一地区疾病流行情况的可视化可以为个人选择提供信息，例如限制在相关时间段内与他人的互动。然而，可视化可能是有缺陷的，这可能导致对数据的误解，并在危机中导致具有负面后果的决策。这个项目试图找出可视化的各个方面，让它们被广泛分享，找出可视化可能存在的缺陷，并警告社交媒体用户。最终，这个项目可以促使普通民众对危机做出更好的反应，并有助于提高可视化素养。最后，该项目还将培训两名研究生，为本科生研究提供机会，并为教授可视化设计的教育工作者提供可利用的材料。这些目标将通过将现有的和新的方法，如主题建模和社会关注度计算方法，应用于与最近的危机相关的三个社交媒体帖子的大型数据集来实现。使用定性编码方法，将开发设计问题的分类。此分类法将用于标记大型数据集。最后，将开发一个插件形式的原型干预，它警告可视化有问题，但也使用户能够根据他们遇到的可视化对问题进行分类。在此项目过程中编译的数据集和注释将公开共享。创建的软件将在许可的、非病毒的开源许可证下发布。该奖项反映了NSF的法定使命，并已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。

In a time of crisis, such as during a hurricane or a global pandemic, social media is an important source of information for the general population. In these scenarios, data visualizations are often used to convey information that is critical for decision making by individuals. For example, a visualization of the path of a hurricane can inform the affected population about the need to prepare or evacuate; while a visualization about the prevalence of a disease in a certain area can inform personal choices, such as limiting interactions with others during a relevant time period. Visualizations, however, can be flawed, which can lead to misinterpretation of the data, and, in a crisis, lead to decisions with negative consequences. This project seeks to identify aspects of visualizations that makes them widely shared, identify flaws a visualization might have, and warn social media users about them. Ultimately, this project can lead to better responses to a crisis by the general population, and contribute to improving visualization literacy. Finally, this project will also enable the training of two graduate students, provide opportunities for undergraduate research, and curate material that can be leveraged by educators teaching about visualization design.These goals will be achieved by applying existing and novel methods, such as topic modeling and calculating measures of social attention, to three large dataset of social media posts related to recent crisis. Using a qualitative coding approach, a taxonomy of design problems will be developed. This taxonomy will be used to label a large dataset. Finally, a prototype intervention in the form of a plug-in that warns of problematic visualizations, but also enables users to classify problems with visualizations they encounter, will be developed. The dataset and the annotations compiled in the course of this project will be shared publicly. The software created will be released under a permissive, non-viral open source license.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.