Peace Science Infrastructure (PSI) is developing automated tools for data collection from news sources regarding the onset and escalatory dynamics of war and other forms of armed conflict such as terrorism. PSI will also develop common standards for coding subnational geo-referenced data, develop machine-coding, and machine-learning tools. The general framework of event (location and timing) and actor relation detection are crucial tools for understanding many aspects of the world, including armed conflicts, terrorism, financial markets, disease mapping, and much more.
PSI will create a consistent industry standard infrastructure for collection and management of conflict-relevant event data – through work on three scientific goals: common standards for sub-national geo-refenced data; machine-coding tools; and web platform for open dissemination. The result will be radically increased accuracy, speed, and cost-saving in data coding and analysis, enabling more nuanced and politically relevant knowledge production. We guarantee the PSI infrastructure to be open, transparent, and free.
PSI will establish and consolidate a lasting and fruitful relationship between leading social science research environments in Norway at UiO and PRIO. To this end, the Peace Science Infrastructure (PSI) project brings together specialists in peace science, informatics and natural language processing, and machine learning.
This collaboration ensures that the infrastructure developed is useful for applied research in the social sciences and that the tools developed are based on cutting edge research from computer science, machine learning, and data science, which will benefit the natural sciences. Political scientists, especially peace scientists, will benefit tremendously.
The Peace Science Infrastructure (PSI) project will provide a comprehensive research infrastructure for the next generation of social science using event data, especially peace science. PSI rectifies two barriers to further scientific progress in the field. First, emerging disaggregated data projects lack a common system for actor identification and spatiotemporal data structure, resulting in inconsistent and partly incompatible datasets. Second, demands for more detailed and accurate data on a wide range of political events are increasingly incompatible with existing rigid and costly human-based systems of data coding.
PSI has three scientific goals:
(1) Develop common industry standards for structuring event data in space and time. A crucial component of this work is establishing a common ID framework for non-state actors and entities at various scales, including sub-national.
(2) Develop adaptable machine-coding tools to accurately record social and political events compatible with these entities and actors. This will serve as a critical quality-control mechanism for existing data projects; in the longer run this technology will replace costly and time-consuming human coding.
(3) Develop a web-based platform for open dissemination of codes, routines, and data, as well as research results.
Through the accomplishment of these three goals, PSI will remedy inconsistencies in extant data collection, dramatically reduce costs and increase quality in future data collection, and close the analytical gap between knowledge needs and knowledge production.
PSI will solidify PRIO's leading role in international peace and conflict research. Nationally, PSI will deepen existing ties with the University of Oslo - Political Science and Informatics. Internationally, PSI will strengthen existing ties to the to Uppsala University and SIPRI.