<![CDATA[This study aims to predict the winners of the 2025 National Science Olympiad (OSN) using the Dynamic Bipartite Graph Embedding (DBGE) method. OSN is an annual competition organized by the Ministry of Education, Culture, Research, and Technology of the Republic of Indonesia to develop students' potential in various fields of science and social sciences. Predicting OSN winners is important for strategic planning, allocating human resources, and improving the quality of education. This study models data on OSN participants and winners from 2009 to 2024 in the form of a dynamic bipartite graph. The modified DBGE method is used to predict links, with additional logistic regression to predict the relationship between schools and OSN medals. The results show that the modified DBGE method has an accuracy of 84%, precision of 83%, recall of 86%, F1 score of 84%, and AUC of 0.92, which are better than the initial DBGE method. The prediction results show that several schools have a high probability of winning gold medals in various OSN categories. Schools such as SMA Kristen BPK Penabur Gading Serpong, SMAN 8 Jakarta, and SMA Semesta show dominance in several categories. This prediction can help in strategic planning and resource allocation to improve the quality of education in these schools. This study contributes by modifying the existing DBGE method so that there is an increase in accuracy. In addition, researchers added a local dataset taken from real cases to prove that this method can be applied and utilized in real terms. This method is expected to be a tool that helps in strategic planning to improve the quality of school education in various regions, especially in preparing students for future science competitions.]]>
