About the seminar:
"When and where will violent conflict break out?" The answer to this question is critically important to people who might fall victim to violence, to policy makers who are charged with preventing and resolving deadly disputes, and to academics who strive to understand human behavior. The social science literature on reducing political violence mostly concerns answering the question, "Which countries are likely to experience violence?" Recent research has refined geographic predictions to identify local level danger zones. However, little progress has been made toward the temporally fine-grained question, "On which specific data will violence occur in a specific place?"
Please join us for a lunchtime seminar led by Dr. Sera Linardi on the work she and her colleagues hope to lead to breakthroughs in our current limited understanding of local level violence, to better address its effects on intergroup relations and potential escalations into national-level problems.
Dr. Linardi will be discussing the results of combining violence data from UN peace operations' weekly logs with daily antenna transmission data from Orange Telecom, which have shown a pattern of increased call volume, more within-network calls, and shorter calls in the days preceding violent events.
An R library, developed by Dr. Linardi and student researcher Lujing Li, which uses various machine learning techniques for prediction and plotting the output on GIS maps, will also be discussed during this seminar.
About the speaker:
Dr. Linardi is an assistant professor at the Graduate School of Public and International Affairs (GSPIA) at the University of Pittsburgh with a secondary appointment at the Department of Economics. She received her PhD in Social Science at the California Institute of Technology (Caltech) in 2010. Before Caltech she was a computer scientist at Adobe Systems, working on the PDF file technology.
Her research is motivated by practical problems faced by organizations delivering human /social services. Her work can be divided into three areas: prosocial behavior, information aggregation, and the behavior of social service clients.