About the project
The Research Council of Norway funded Peace Science Infrastructure-project represents one of the most ambitious infrastructure investments ever within social science in Norway. The project is a collaboration between PRIO, Uppsala University, SIPRI, and University of Oslo (Department of Informatics and Department of Political Science).
It will allow social scientists throughout Norway and around the globe to make use of new and unexplored data sources to study in fine-grain detail the dynamics of political processes. The infrastructure developed is broadly applicable to any area of research that attempts to understand the behavior of political actors.
To date, the most widely used datasets in social science register information at the country-year level of analysis. This is problematic. Country-year analysis fails to account for local effects and dynamics. Misidentification also occurs, such as inferring that diamond-fueled armed conflict in Russia because the same sovereign state has both significant diamond deposits (in Siberia) and a civil war (Chechnya). Often, theory is at the micro-level, thus introducing a disturbing gap between micro-level theories and macro-level analysis.
Objectives
Our aim is to develop an infrastructure for automated event coding from news sources. At the core of the system is a large language model that incorporates domain knowledge about political systems and conflict processes. Users of the infrastructure take advantage of the efficiencies of machine learning and execute data collection projects hitherto infeasible.
The project will also standardize definitions of actors, locations, events, and actions within a particular field, which in turn can be used and expanded by third party users. An increased interdependency between data projects will in turn lead to increased quality and collaboration. Or, in other words, better research for less money.
