In 2024, one in 67 people on the planet had been forced to flee their homes. This shocking number is a clear symptom of the increasing global crisis of forced displacement. Our latest study demonstrates the potential of social media sentiment analysis for predicting displacement trends during an emergency, improving humanitarian response. Marahrens, who is an assistant professor of computational social science at Notre Dame’s Keough School of Global Affairs, led this study. It opens up a window into the struggle among displacement mandated by conflict and economic chaos.
The full report is now available in EPJ Data Science. It focuses on three case studies to illustrate the deep and permanent displacement these different crises have caused. In Sudan, an estimated 12.8 million people were already displaced after a civil war broke out in April 2023. Likewise, at least 10.6 million people were forced to leave their homes inside and outside of Ukraine after Russia invaded in 2022. Meanwhile, Venezuela continues to face an untold tragedy. Over the past few years, tens of millions have been displaced by a string of brutal recessions.
In just the past 10 years, the worldwide displaced population has nearly doubled. This troubling data comes from the United Nations’ refugee agency. The surge in forced displacement has created a global crisis that necessitates improved methods for tracking and responding to migration patterns.
Marahrens and his team delved deep into almost 2 million social media posts in three languages across X (formerly Twitter). From this, their goal was to determine how online sentiment reflects what is happening in the real world with displacement events. The findings indicate that social media analysis is particularly effective in conflict situations, such as those experienced in Ukraine, where immediate threats prompt rapid relocations.
What we found in our research was a surprising juxtaposition. Social media sentiment analysis has a hard time during economic crises, like in Venezuela, as these situations often need time to build up. This difference suggests that the nature of a crisis significantly influences how people communicate their experiences and decisions on social media platforms.
Marahrens noted that more research in the future could explore the connections between sentiment and emotion. Our overarching research question is how these different worlds mesh and clash at times of crisis. Through improved comprehension of these interactions, researchers can create more effective instruments to assist with humanitarian endeavors.
This study is not just pouring it over some raw data. It starkly illustrates the ongoing need for more tailored, rapid response mechanisms that are attuned to the specific contexts of displacement. Through the power of social media sentiment analysis, humanitarian organizations will be able to more accurately predict movements of at-risk populations and distribute resources accordingly.