언론 논조를 통해 예측한 미국 대통령선거

초록

This study proposes a novel approach that combines media sentiment analysis with big data to predict the outcome of the U.S. presidential election. Using data from the GDELT project, it quantitatively analyzes national and state-level media sentiment regarding candidates Kamala Harris and Donald Trump to forecast election results. Specifically, the study builds upon the existing theory that media sentiment can serve as a leading indicator of voter preferences to predict support trends at both the national and state levels. The research employs a Multivariate LSTM deep learning model to account for the interdependence between media sentiment and election outcomes. The data is updated daily, enabling real-time predictions. The analysis reveals that competition between the two candidates in the 2024 election remains intense. While Harris shows a disadvantage at the national level, outcomes in swing states remain highly uncertain. Beyond election forecasting, this study also analyzes the policy priorities of candidates through media coverage, providing critical insights for South Korea’s foreign and domestic policy planning. This methodology is expected to be applicable to forecasting and policy analysis for future South Korean presidential elections.

키워드