EAGER: SaTC: Shifts in Misinformation Topics on Social Media: Manipulators Masquerading as Humans

_EAGER：SaTC：CORE：Small：社交媒体上错误信息主题的转变：伪装成人类的操纵者社交媒体上错误信息的传播可能会对公众的健康、福祉和稳定造成重大后果。从医学错误信息到政治错误信息，各种各样的话题都容易受到错误信息的影响。传播错误信息的账户可以大致分为两类：（1）无意传播错误信息的账户（即，相信他们传播的错误信息的个人）和（2）以故意欺骗为目的这样做的人（即，虚假信息的代理人“伪装”成人类）。前一类人通常在有限数量的主题上传播错误信息（即，无论是医学还是政治，但不是两者兼而有之），专注于他们作为个人所关心的事情。然而，恶意的第三方行为者可能会激励虚假信息的代理人在不受限制的各种主题中传播错误信息，目的是在公众中引发广泛的不稳定。该项目分析了社交媒体上传播的错误信息，以区分第三方激励的虚假信息代理人和其他更良性的账户。为了实现这一目标，该团队将研究Twitter的数据，以确定2022年上半年在传播医疗错误信息和传播政治错误信息之间快速切换的账户。一个机器学习框架将被设计成从这个账户子集所特有的语言特征中学习，然后将被用来开发一个分类工具，在更广泛的Twitter领域（即，2022年前和2022年后）作为“潜在的虚假信息代理人”。该团队还将描述2022年上半年传播的错误信息中有多大部分可归因于此类第三方激励代理。该奖项反映了NSF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。

__________________________________________________________________________________________________________________________________EAGER: SaTC: CORE: Small: Shifts in misinformation topics on social media: manipulators masquerading as humansThe spread of misinformation on social media can result in major consequences to the health, wellbeing, and stability of the general public. A wide range of topics are vulnerable to misinformation, varying from medical misinformation to political misinformation. Accounts that spread misinformation can be broadly classified into two categories: (1) those who do so unintentionally (i.e., individuals who believe in the misinformation that they spread) and (2) those who do so with the aim of being deliberately deceptive (i.e., agents of disinformation “masquerading” as humans). Those in the former category typically spread misinformation on a constrained number of topics (i.e., either medical or political, but not both), focusing on what they care about as individuals. However, agents of disinformation may be incentivized by malicious third-party actors to spread misinformation across an unconstrained variety of topics, with the objective of prompting widespread instability among the general public. This project analyzes misinformation spread on social media to distinguish third-party-incentivized agents of disinformation from other, more benign accounts.To achieve this goal, the team will examine data from Twitter to identify accounts that switched rapidly between spreading medical misinformation to spreading political misinformation during the first half of 2022. A machine learning framework will be designed to learn from linguistic features that are unique to this subset of accounts, which will then be used to develop a classification tool to label accounts across the broader Twittersphere (i.e., pre-2022 and post-2022) as “potential agents of disinformation”. The team will also characterize what fraction of misinformation spread during the first half of 2022 was attributable to such third-party-incentivized agents. All algorithms developed over the course of the project will be shared openly with the broader scientific community to facilitate efforts towards countering disinformation on social media.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.