filmov
tv
Project Domino: Citizen data science against COVID misinformation
Показать описание
With elected US leadership now openly promoting medical misinformation and conspiracy theories, over 150,000 Americans have died from COVID-related causes. Behind the scenes, misinformation groups are successfully amplifying their social media campaigns by cooperating with one another, national political parties, and digital media monopolies. This talk shares the perspective from Project Domino, an effort to power anti-misinformation interventions by taking an open citizen science approach. First, we will provide a security analytics view of the operations by misinformers, including how they collaborate. Second, we discuss technical and systemic gaps that misinformers are exploiting, and where open source citizen science fits in. Finally, we share some of the enabling technologies for open misinformation data science (DIY tech around Jupyter, StreamLit, RAPIDS, Graphistry, …) and emerging relevant data methods (BERT/FAISS, UMAP, GPU graph analytics, …). As a result, we have been able to help misinformation researchers and reporters tackle higher scales and get better results. Surprisingly, we are optimistic as we layer on tools for use by regular citizens, leaders of at-risk communities, and other organizations.
Presented at AI Village @ DEF CON 28, safemode
10am Saturday August 8th, 2020
Presented at AI Village @ DEF CON 28, safemode
10am Saturday August 8th, 2020
Комментарии