CAREER: Toward Equitable Social Media Content Moderation for Marginalized Individuals and Communities

本研究考察了在线平台禁止用户和内容的原因和后果，这些用户和内容实际上并没有违反网站政策，或者处于网站政策和社区规范的灰色地带。它将实现：(1)描述人们被禁止或从社交媒体网站上删除内容的原因的类型和普遍性，这些与系统性偏见的关系，以及这些做法的影响。(2)深入描述边缘化在线社区为解决适度导致的不平等而制定的流程和支持结构。(3)结合定性和计算技术方法，根据社区规范和不断变化的网站政策，更准确地划分应该和不应该被禁止的社交媒体内容。(4)为上下文感知的内容审核系统设计建议，该系统将社区需求、规范和价值观与平台要求相平衡，并可应用于所有社交媒体网站。(5)关于当前在线内容审核不公平的理论见解，以及如何通过承认和尊重边缘化在线社区的需求来更公平地实现审核目标。将进行以下研究：1)访谈被边缘化的人，他们的内容或账户被从社交媒体网站上删除，即使符合网站政策或处于灰色地带；2)三个边缘化网络社区的虚拟人种志，这些社区经常成为用户封杀和内容删除的目标。3)使用参与式设计和焦点小组方法来确定如何建立一个可应用于跨社交媒体网站的上下文感知的内容审核系统。4)在线实施该社会技术系统，并进行评估以了解用户认为它公平的程度以及它如何满足他们的需求。此外，数字素养在线资源将帮助人们了解内容和帐户删除以及如何重新获得网站访问权，研究结果将使社交媒体平台了解边缘化人群在内容审核方面面临的挑战。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 research examines the causes and consequences of online platforms banning users and content that do not actually violate site policies, or fall into gray areas with respect to site policies and community norms. It will achieve: (1) A characterization of the types and prevalence of reasons people are banned or have content removed from social media sites, how these relate to systemic biases, and the implications of these practices. (2) An in-depth description of processes and support structures marginalized online communities enact to address inequities resulting from moderation. (3) A combined qualitative and computational technical approach that more accurately delineates between social media content that should and should not have been banned, according to community norms and shifting site policies. (4) Design recommendations for a context-aware content moderation system that balances community needs, norms, and values with platform requirements and can be applied across social media sites. (5) Theoretical insights around current inequities in online content moderation and how moderation goals can be achieved more equitably by recognizing and respecting marginalized online communities’ needs. The following studies will be conducted: 1) Interviews with marginalized people whose content or account was removed from a social media site even though in line with site policies or falling into a gray area; 2) A virtual ethnography in three marginalized online communities who are frequent targets of user bans and content takedowns. 3) Use of participatory design and focus group methods to determine how to build a context-aware content moderation system that can be applied across social media sites. 4) Implementation of this sociotechnical system online, and evaluation to understand the extent to which users consider it equitable and how well it meets their needs. In addition, a digital literacy online resource will help educate people about content and account removals and how to regain site access, and the results of the research will educate social media platforms about challenges marginalized populations face around content moderation.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.