Social media has become a primary platform for the public to express their opinions. Since 2023, politician XYZ has been one of the most widely discussed figures, particularly on Social Media X. Several political events between 2023 and 2024 make the public sentiment toward this figure interesting to analyze. This study conducts a temporal sentiment analysis of public opinion on politician XYZ from August 2023 to March 2024, using FastText word embeddings and a Graph Neural Network (GNN) model. The approach involves data collection, text processing, and sentiment classification, utilizing FastText to capture the semantic relationships between words and a Graph Neural Network (GNN) to model sentiment dynamics over time. The focus of this study is to explore the temporal aspect of sentiment shifts, providing insights into how public opinion evolves over time in response to political events, in contrast to static sentiment. The temporal sentiment analysis reveals that the public’s perception of politician XYZ initially began with positive sentiment but shifted to negative sentiment in the following months, influenced by key political events. With an accuracy of 72%, this study highlights the potential of integrating FastText and GNN for capturing complex and evolving political sentiments. The findings offer practical implications for political communication strategies, enabling stakeholders to better understand and anticipate shifts in public opinion during critical political moments.