Racism in social media is ubiquitous, persisting online in ways unique to the internet while also reverberating from the world offline. When will racist frames activate in social media networks? This article argues that social media users engage with racist content when they perceive a threat to the in-group status, selecting frames that serve as markers to separate the in-group identity from the out-group identity. Racialized frames serve as these markers, and the perceived threats to the in-group status make racist content cognitively congruent. Evidence of this behavior is provided by examining Twitter activity during the indígena protests in Ecuador in October 2019. A novel, multistep machine-learning process detects racist tweets, and an interrupted time series analysis shows how events that can be perceived as threats to the in-group activate racist content in some social media communities.

中文翻译：

---

机器之怒：社交媒体中种族主义内容的激活

社交媒体中的种族主义无处不在，以互联网独有的方式在网上持续存在，同时也在线下世界中产生反响。种族主义框架何时会在社交媒体网络中激活？本文认为，社交媒体用户在感知到对群体内地位的威胁时会参与种族主义内容，选择作为标记的框架来区分群体内身份和群体外身份。种族化的框架充当这些标记，对群体内地位的感知威胁使种族主义内容在认知上是一致的。通过检查 2019 年 10 月厄瓜多尔土著抗议期间的 Twitter 活动，可以提供这种行为的证据。一种新颖的多步骤机器学习过程可以检测到种族主义推文，