CAREER: Developing Computational Methods to predict Hate Crimes

根深蒂固的群体信仰会激励人们从事仇恨犯罪。然而，预测这些犯罪是极具挑战性的，因为它们在统计上非常罕见。该项目使用计算方法和社交媒体上的大数据来预测恶毒的仇恨言论何时会转变为仇恨犯罪。这项研究考察了仇恨言论的语言及其在社交网络上的传播如何预测仇恨犯罪。其科学目标是提供信息并促进对极端和仇恨思想背后的社会和认知过程以及它们如何传播的理论和基本理解。其中一个目标是确定可能阻止仇恨言论在社交媒体上传播的机制。另一个是为执法提供新的工具。许多学科的学生和教职员工将有机会学习处理大数据的新计算方法，以及检查在线和通过社交媒体传播思想的方法。通过将社会认知理论与自然语言处理和机器学习技术相结合，该项目试图了解根深蒂固的信念和身份动态是如何结合在一起形成极端观点的，从而可能导致贬低外部群体和仇恨犯罪。它考察了某些形式的社会认知和仇恨言论的语言特征，并考虑了如何从稀疏的、有偏见的数据中估计仇恨言论的空间分布。这项研究考虑了是否可以使用不引人注目的措施和关于社交网络的信息来预测社会动机的仇恨犯罪和仇恨言论行为。相关的教育和培训机会包括指导少数群体服务机构的学生，与更广泛的科学界分享新开发的文本分析工具，以及关于机器学习和文本分析的讲习班。该职业奖项由NSF社会心理学和安全可信网络空间(SATC)项目共同支持。该奖项反映了NSF的法定使命，并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。

Deeply-held group beliefs can motivate people to engage in hate crimes. Yet predicting these crimes is extremely challenging because of their statistical rarity. This project uses computational methods and big data from social media to predict when virulent hate speech transitions to hate crimes. The research examines how the language of hateful speech and its dissemination over social networks predicts hate crimes. The scientific goal is to inform and advance theory and basic understanding of the social and cognitive processes that underlie extreme and hateful thoughts and how they spread. One aim is to identify mechanisms that might inhibit the dissemination of hate speech in social media. Another is to provide new tools for law enforcement. Students and faculty across many disciplines will have an opportunity to learn new computational methods for handling big data and for examining the spread of ideas online and through social media. By integrating theories of social cognition with natural language processing and machine learning techniques, this project seeks to understand how deeply-held beliefs and identity dynamics coalesce to form extreme perspectives that could result in derogation of out-groups and hate crimes. It examines the linguistic features that characterize certain forms of social cognition and hate speech, and considers how the spatial distribution of hate speech can be estimated from sparse, biased data. The research considers whether socially-motivated hate crimes and acts of hate speech can be predicted using unobtrusive measures and information about social networks. The related educational and training opportunities include mentoring of students from minority-serving institutions, sharing newly-developed tools for text analysis with the wider scientific community, and workshops on machine learning and text analysis. This CAREER award is supported jointly by the NSF Social Psychology and Secure and Trustworthy Cyberspace (SaTC) Programs.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.