We will develop accurate and robust techniques for extracting and summarizing information about events and beliefs from free text. The center will focus on three areas of research. (1) We will create easily trainable learning algorithms that can automatically create domain-specific patterns to identify facts and relations associated with relevant events, such as infectious disease outbreaks. (2) We will develop trainable learning algorithms that can distinguish factual assertions from subjective (non-factual) assertions, identify beliefs that are held by an entity, and assess the intensity, polarity, and motivation and attitude types of those beliefs. (3) We will create methods for understanding event and belief progressions over time.

This project is funded by the Department of Homeland Security as a University Affiliate Research Center (UAC). The purpose of the UAC program is to fund basic research and education, and to foster research collaborations among universities and the National Laboratories. The other UACs are at Rutgers University, the University of Southern California, and the University of Illinois at Urbana-Champaign.