Accurate situational awareness becomes an increasingly difficult challenge in rapidly changing environments. With currently exponential growth of COVID-19 confirmed cases, timely and reliable information becomes extremely important for informed decision making. Official reports based on confirmed test results are reliable, but widely considered to be a subset of the real situation. In contrast, social media provide broad coverage, but they have low reliability due to significant misinformation and disinformation or inaccurate news. With the gradual opening of businesses in the US, while the prospect of an effective vaccine remains uncertain, the need for reliable and accurate situation awareness becomes paramount, since the decisions for further business openings and practices of social distancing will depend on the information and perception of risks of contagion and the need for economic recovery.
This project addresses the technical challenges of finding new, verifiable facts from noisy online media and social networks in a timely manner. Social media contain the necessary timely information, but they also carry significant challenges represented by misinformation, disinformation, and concept drift. Traditional machine learning (ML) models trained from closed data sets have been unable to meet these challenges when faced with true novelty in evolving new data, beyond the fixed training data. To handle these challenges, the Evidence-Based Knowledge Acquisition (EBKA) approach automates the integration of noisy social media data such as Twitter and Weibo with recognized, respected authoritative sources to detect verifiable facts timely and reliably. The project build on the LITMUS software tools to provide timely and reliable information com complement physical test result data, and enable better informed decision making by government officials, first responders, and the general public.
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.
