Sentiment Analysis of International Relations with Artificial Intelligence

Abstract

Geopolitical strategy is characterized by a dynamic and complex structure of entity relationships, geo-spatial data and human decisions. We employ machine and deep learning techniques to retrieve the sentiment between countries through scraping and analyzing news articles. The change in the sentiment score between countries allows to analyze historic developments of international relations as well as to evaluate the primary and secondary network effects of potential events and policy decisions on the global relationship structure. We find that the key for the most accurate real mapping of the sentiment score between countries is the maximization of the quantity of news while simultaneous minimization of the noise added by the news. Moreover, we show the potential of Artificial Intelligence (AI) to improve and forecast international relations. Keywords: Natural language processing, international relations, sentiment analysis, geo-political forecasting.