[论文解读] Racism in the Machine: Visualization Ethics in Digital Humanities Projects

Katherine Hepworth, Christopher M. Church|arXiv (Cornell University)|Nov 12, 2024

Digital Games and Media被引用 15

一句话总结

论文认为可视化本质上具有修辞性和偏见，提出一个伦理可视化工作流，并以两个数字映射项目进行说明。

ABSTRACT

Data visualizations are inherently rhetorical, and therefore bias-laden visual artifacts that contain both explicit and implicit arguments. The implicit arguments depicted in data visualizations are the net result of many seemingly minor decisions about data and design from inception of a research project through to final publication of the visualization. Data workflow, selected visualization formats, and individual design decisions made within those formats all frame and direct the possible range of interpretation, and the potential for harm of any data visualization. Considering this, it is imperative that we take an ethical approach to the creation and use of data visualizations. Therefore, we have suggested an ethical data visualization workflow with the dual aim of minimizing harm to the subjects of our study and the audiences viewing our visualization, while also maximizing the explanatory capacity and effectiveness of the visualization itself. To explain this ethical data visualization workflow, we examine two recent digital mapping projects, Racial Terror Lynchings and Map of White Supremacy Mob Violence.

研究动机与目标

• 激发数字人文学中对数据可视化的伦理问责。
• 指出数据工作流和设计选择如何塑造解读与潜在伤害。
• 提出一个伦理数据可视化工作流，在最大化解释力的同时最小化伤害。

提出的方法

• 阐明可视化包含明确和隐含的论证。
• 分析数据工作流、可视化格式和设计决策如何影响解读。
• 提出一个伦理数据可视化工作流，在最小化伤害与提升解释效果之间取得平衡。
• 检视两个案例研究：Racial Terror Lynchings 和 Map of White Supremacy Mob Violence.

实验结果

研究问题

• RQ1 数据工作流和设计选择在可视化中如何影响解读和潜在伤害？
• RQ2 在数字人文学中，创建和使用数据可视化的伦理工作流应包含哪些要素？
• RQ3 如何在实际的地图项目中将伦理考量整合到实践中？

主要发现

• 可视化从数据选择到最终设计的决策形塑出隐含论证。
• 伦理可视化实践可以在保持解释力的同时减少对研究对象和受众的伤害。
• 用两个数字映射项目来说明所提议的伦理工作流：Racial Terror Lynchings 和 Map of White Supremacy Mob Violence.