Collaborative Research: SaTC: CORE: Medium: Information Integrity: A User-centric Intervention

在数字时代打击虚假信息是一项具有重大社会影响的挑战性课题，因为虚假信息继续影响着从选举到应对流行病等有争议的当代事件。尽管经过数十年的研究，错误信息仍然是一个严重的威胁，因为大多数技术缓解方法侧重于提高检测准确性，而未能考虑社会和情感角度。本项目通过确定有影响力的社区、代理人和具有文化共鸣的信息，以确定有关争议问题的公共对话的转折点，并提供以用户为中心进行大规模干预的场所，帮助加强信息完整性。该项目从以源为中心的准确性检测和揭穿转向以用户为中心的干预，将心理和社会文化结构、计算理论和机器学习（ML）算法集成到原型干预测试中。研究的第一个重点是分析和识别社会规范的出现——社区的共同信念或可接受的行为，以及信念即将迅速改变的临界点。本研究的第二个重点是揭示信仰的文化背景，个性化到每个个体，以优化在线网络传播中科学证据的可接受性。第三个支柱（交互）提供了人在循环中的可视化分析框架，以支持用户验证并根据他们的信念做出用户自己的决定。这项工作的基础是基于拓扑机器学习的新型深度学习模型的开发和测试，该模型有效地预测异质社会规范的出现，以便及时干预，识别参与的顶级可信特征，以及与用户透明交互的时间可解释的人工智能。波因特研究所等主要的减少错误信息和新闻教育组织的参与有助于确保在该领域产生社会影响。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。

Combating misinformation in the digital age has been a challenging subject with significant social implications, as misinformation continues to impact contentious contemporary events from elections to responses to pandemics. Despite decades of research, misinformation remains a serious threat as most technical mitigation methods focus on improving detection accuracy and fail to consider social and emotional perspectives. This project assists in enhancing information integrity by identifying influencing communities, agents, and culturally resonant information to identify tipping points in public dialogue on controversial issues and offering venues of user-centric interventions at scale.This project moves away from source-centric accuracy detection and debunking to focus on user-centric interventions that integrates psychological and socio-cultural constructs, computational theories, and machine learning (ML) algorithms to prototype interventions for testing. The first focus of research has the goal of analyzing and identifying social norm emergence--the shared beliefs or acceptable behaviors of communities, and tipping points when beliefs are about to change rapidly. The second focus of this research is to uncover the cultural contexts of belief, personalized to each individual, to optimzie the receptivity of scientific evidence in online network dissemination. The third pillar (Interaction) provides human-in-the-loop visual analytics framework to support users in verifying and making users' own decisions as to what they belief. Underpinning this work is the development and testing of novel deep learning models based on topology ML, which effectively predict heterogeneous social norm emergence for timely intervention, identify top trusted features for engagement, and temporal explainable artificial intelligence for transparent interaction with users. The involvement of leading misinformation mitigation and journalism education organizations such as the Poynter Institute helps to ensure social impacts in the field.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.