This paper explores international event forecasting using GDELT:
International event data can be seen as sensors that explain the interaction of countries. With the help of international event data, we can quantitatively analyze bilateral and even multilateral relations, which has very important reference value for stakeholders such as multinational companies and policy makers. This article explores the potential of the GDELT large database in the analysis and prediction of international event data. First, we introduced the GDELT database, data format and coding system of International Events Data Analysis. Second, we select all interaction events between China and other countries in the world from February 2020 to September 2021 to understand China's interaction with other countries and the trend of China's international evaluation through the analysis of conflict-cooperative events and media tone. Finally, we selected a more random "AvgTone" field for prediction, and proposed a prediction model based on the Encoder-Decoder Attention framework. After experiments, the model can still converge against data with a lot of noise, which proves the potential value of deep learning algorithms in international event data analysis.