EAGER: Data Science for Election Verification

点击翻译按钮获取中文摘要

Election officials need evidence-based, scientifically valid tools to routinely assess the quality of election systems, including technical and human factors. Whether initiated by the election administrators or by the parties to an election, election investigation is expensive and must be well prioritized to be most effective. The project plans to provide tools for prioritizing investigations for election officials and others. The expected outcome is increased robustness of the mechanisms protecting elections that can give Americans more justified confidence in election results.The research will help transform the way data science is used to protect elections. Society does not yet have a sustainable ecosystem of systematic data science to inform decisions by election administrators, political parties and candidates about how to prioritize their limited investigation resources. The peculiarities of elections may require development of new techniques in predictive analytics. The goal is to predict the behavior of a collection of elements, not the behavior of single elements. The analysis must be done in the presence of local noise and overall trends created by idiosyncratic effects of candidates, campaigns and cultural trends.The project's goal is to bolster the actual and perceived integrity of American elections. Election systems are critical infrastructure. Protecting elections is part of national security. This will increase the resilience of US election infrastructure by improving our ability to detect and correct erroneous election results. In addition, because of the high visibility of election security issues, the work will promote public engagement with data science.The research will collect data on the technology fielded by election agencies across the country. This data will be publicly available at www.VerifiedVoting.org/verifier. The research will also create prototype analysis and visualization tools that will be shared publicly. The research will progress from relatively simple analyses that have had a demonstrable impact on real elections in a few jurisdictions to analyses involving a wider variety of data, including census and election technology data, and more sophisticated predictive analytics and machine learning algorithms.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.