EAGER: DCL: SaTC: Enabling Interdisciplinary Collaboration: Combatting Disinformation and Racial Bias: A Deep-Learning-Assisted Investigation of Temporal Dynamics of Disinformation

这个项目探讨了种族虚假信息在网上的传播及其社会影响，特别关注亚裔美国人。在新冠肺炎疫情期间，针对亚裔美国人的仇恨和偏见变得明显，但针对亚裔的虚假信息尚未得到充分挖掘。该项目的新颖之处在于采用独特的多学科方法：(1)检测针对亚洲的虚假信息及其对策信息，并了解它们如何在网络上传播；(2)研究网络上虚假信息和对策信息的传播如何与针对亚裔美国人的偏见和仇恨犯罪的强度相关联；(3)开发各种数据驱动的计算模型，以帮助理解虚假信息的动态。该项目更广泛的意义和重要性在于，使包括倡导组织和公众在内的民间社会了解如何制定打击种族虚假信息的传播战略，并将开发的工具和成果公之于众，供更广泛的用途。该项目采用三阶段方法。第一阶段的主要目标是开发计算工具，用于检测和分析网络上针对亚洲的虚假信息和对策之间的时间动态。具体的重点是通过特征工程的独特特征来开发自动识别工具和深度学习分类模型。第二阶段的目标是了解网上虚假信息和对策的传播在多大程度上与现实世界中针对亚裔美国人的隐性偏见和仇恨犯罪事件的社会趋势有关，这可以通过开发深度学习因果关系模型来实现。第三阶段的目标是在宏观和微观层面设计可扩展的数据驱动的虚假信息动态深度学习模型，从现实世界的测量中识别未知动态，这也使得学习动态的模拟成为可能。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。

This project explores the diffusion of racial disinformation online and its social impacts, particularly focusing on Asian Americans. While the hatred and bias against Asian Americans have become notable amid the COVID-19 pandemic, Asian-targeting disinformation has yet been fully explored. The project's novelties are in unique multidisciplinary approaches to (1) detect Asian-targeting disinformation and its countermeasure messages, and understand how they are spread on the web, (2) examine how the spread of disinformation and countermeasure messages on the web is associated with the intensity of the bias and hate crimes against Asian Americans, and (3) develop various data-driven computational models to help understanding the disinformation dynamics. The project's broader significance and importance are to inform civil society, including advocacy organizations and the general public, about how to strategize communication efforts in battling racial disinformation, and to make the developed tools and outcomes publicly available for broader uses.The project takes three-staged approaches. The main objective of the first phase is to develop computational tools for the detection and analysis of the temporal dynamics between Asian-targeted disinformation and countermeasures on the Web. A specific focus is on developing automated identification tools and deep-learning classification models by feature-engineering unique characteristics of disinformation. The objective of the second phase is to understand to what extent the spread of disinformation and countermeasures online is associated with the societal trend of implicit bias and hate crime occurrences against Asian Americans in the real-world, which can be achieved via developing deep-learning causality models. The objective of the third phase is to design scalable data-driven deep-learning models of disinformation dynamics in macro and micro levels, identifying unknown dynamics from the real-world measurements, which also enables simulations of the learned dynamics.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.