CRII: SaTC: Towards Understanding and Defending Against New Waves of Online Hate

快速变化的世界事件，如COVID-19大流行，伴随着社交媒体平台上的话语发生了重大变化。网上的仇恨不断增加，并且一波接一波地到来；这是一个深刻关注的问题，正在消极地改变着互联网用户的生活。最近的研究表明，网络仇恨是如何违反社交媒体政策的，并描述了这些违规行为在现实世界中的后果。网络仇恨很难进行科学研究。仇恨传播的稀疏和有偏差的样本可用于计算分析，特别是当仇恨传播突然发生和传播时。这个项目推进了知识的前沿和我们防御和减轻网络仇恨的能力。该项目采用新颖的方法来研究网络仇恨浪潮，使用新的计算方法来检测网络仇恨政策违规行为，并提出了在社交媒体平台上进行审核的新方法。为了实现这些目标，调查正在发现和编目具有新一波网络仇恨特征的新因素。分类过程是基于对在线仇恨传播的时间和社会测量分析。该项目还在制定新技术，以有效地发现社交媒体流中的联系。关键思想是有效地对在线仇恨数据集进行采样，这样只有表征新实例或形式的在线仇恨的样本才能用于机器学习训练。机器学习范式只需要几个样本就可以有效地学习检测新的在线仇恨浪潮。该项目还使用新技术来识别和跟踪不同社交媒体平台用户社区中在线仇恨的跨平台转移。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。

Rapidly changing world events, such as the COVID-19 pandemic, have been accompanied by significant changes in discourse on social media platforms. Online hate has increased and arrives in waves; it is a deeply concerning problem that is negatively transforming the lives of internet users. Recent studies have demonstrated how online hate violates social media policies and describes the ramifications of these violations in the real world. Online hate is difficult to study scientifically. Sparse and biased samples of hate communications are available for computational analysis, especially when hate communications occur and spread suddenly. This project advances the frontiers of knowledge and our ability to defend against abd mitigate online hate. The project applies novel approaches to study waves of online hate, using new computational approaches to detect online hate policy violations and proposing new methods for moderation on social media platforms.To achieve these goals, the investigation is discovering and cataloging novel factors that characterize new waves of online hate. The categorization process is based on temporal and social measurement analyses of online hate communications. The project also is formulating new techniques to effectively discover linkages in social media streams. The key idea is to efficiently sample online hate datasets such that only samples that characterize new instances or forms of online hate are used for machine-learning training. The machine learning paradigm only needs a few samples to effectively learn to detect the new waves of online hate. The project also uses novel techniques to identify and track cross-platform transfers of online hate in user communities across different social media platforms.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.