EAGER: SaTC: Early-Stage Interdisciplinary Collaboration: Collaborative: Advances in Socio-Algorithmic Information Diversity

如今，社交媒体在让人们接触到从娱乐到硬新闻和政治辩论等广泛话题的信息方面发挥着重要作用。在这些平台上可以看到的内容很大程度上受到算法的影响，这些算法旨在为每个用户选择最吸引人、最相关的内容。通过寻求最大程度的参与，这些算法可能会无意中放大事实可疑或质量差的信息，从而强化用户现有的信念。这样做，这些算法可以减少用户暴露的信息的多样性。该项目将开发新的内容推荐算法，以降低这种风险，提高社交媒体上传播的信息的质量和多样性。本研究将发展对耦合的网络-人类系统如何在社交媒体上的新闻消费背景下处理信息的理解。这种情况在社会、行为、认知和算法层面造成了重要的信息处理漏洞。使用来自美国人口的全国代表性样本的数据，调查人员将测量政治态度、读者、参与度和信息质量之间的关系。他们还将测试旨在促进浏览器扩展/智能手机应用程序中多样化信息消费的行为推动的效果。最后，研究人员将开发一个通用建模框架，以评估这些建议对受众倾斜多样化的影响，并测试它们对欺诈性（先令）攻击的稳健性。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。

Social media now play an important role in exposing people to information about a wide range of topics ranging from entertainment to hard news and political debate. What can be seen on these platforms is heavily influenced by algorithms that are designed to select the most engaging and relevant content for each user. By seeking to maximize engagement, these algorithms may inadvertently amplify factually dubious or poor quality information that reinforces users' existing beliefs. In doing so, these algorithms could reduce the diversity of information to which users are exposed. This project will develop new content recommendation algorithms that reduce this risk and improve the quality and diversity of information circulating on social media.This research will develop an understanding of how coupled cyber-human systems process information in the context of news consumption on social media. This context creates important information-processing vulnerabilities at the social, behavioral, cognitive, and algorithmic levels. Using data from a nationally representative sample of the U.S. population, investigators will measure the association between political attitudes, readership, engagement, and information quality. They will also test the effect of behavioral nudges designed to promote the consumption of diverse information in a browser extension/smartphone app. Finally, the researchers will develop a generic modeling framework to evaluate the effect of these recommendations on audience-slant diversification and to test their robustness against fraudulent (shilling) attacks.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.