Chris Rauh, da University of Cambridgevai, participar num seminário de Ecpnomia.

Predicting conflict - An overview

Abstract:

In this project we propose a framework to tackle conflict prevention, an issue which has received interest in several policy areas. A key challenge of conflict forecasting for prevention is that outbreaks of conflict in previously peaceful countries are rare events and therefore hard to predict. To make progress in this hard problem, this project summarizes more than five million newspaper articles using a topic model. The topics are then fed into a random forest to predict conflict risk. We are currently advancing our research agenda in two directions. First, we are predicting conflict not only at the country, but also at the local level while taking into consideration geo-spatial features of conflict. Second, we are integrating our predictions into a framework in which a decision maker decides on the optimal timing of interventions to minimize the total cost of conflict and intervention. This presentation will give an overview of the work in progress.