INTERNATIONAL SENTIMENT ANALYSIS THROUGH ONLINE NEWS MEDIA

Abstract

This thesis leverages vast data resources to understand contemporary conflict dynamics amidst global competition. Focusing on sentiment analysis of online news media, it explores sentiment’s predictive potential for conflict at the interstate level. Traditional explanations in international relations theory—economic ties, shared values, and power differentials—are extended to include nuanced sentiment analysis. Using Poisson regression, the study establishes a statistically significant correlation between media sentiment and conflict likelihood, surpassing traditional explanatory variables. Robustness checks and peace-month subsets reinforce these findings, suggesting sentiment as a leading indicator of future conflict. This thesis not only expands empirical evidence on sentiment’s impact on human behavior but also advocates for its integration into conflict prediction frameworks. With advancing computing capabilities, this study underscores the importance of analyzing specific sentiment components for deeper insights into social discourse. Ultimately, it offers practical implications for policymakers and information practitioners, emphasizing the value of sentiment analysis in informing strategic decisions amidst evolving global dynamics.