Researchers say analysis of social media posts can give an early warning about when and where people move in crises. This could help humanitarian agencies respond faster when surveys and field data are hard to collect.
The study appears in EPJ Data Science. The team looked at nearly 2 million posts in three languages on X. They studied three cases: Ukraine, Sudan and Venezuela. Ukraine saw 10.6 million people displaced, Sudan about 12.8 million, and Venezuela about 7 million.
Researchers found that sentiment labels (positive, negative, neutral) were more reliable than emotion labels for predicting movements. Pretrained language models gave the most useful early warning. The authors say the method works best in conflict settings and should be used with traditional data to avoid false alarms.
Difficult words
- predict — To say what will happen in the future.
- displaced — People who have been forced to leave their home.
- information — Facts or details about something.
- important — Having great significance or value.
- traditional — Relating to customs or practices from the past.
Tip: hover, focus or tap highlighted words in the article to see quick definitions while you read or listen.
Discussion questions
- How can social media improve aid delivery during crises?
- What other tools might help in similar situations?
- Why is it important to provide aid to displaced people?
Related articles
Brothers build magnetic system to remove arsenic
Arsenic in Indian groundwater causes serious health problems. Two brothers from Bihar developed METAL, a chemical-free magnetic way to clean water and built the MARU unit; their startup Navmarg has treated over 300,000 litres and plans sensors and AI.
Gum ingredients help tilapia cope with cold
Researchers tested lecithin and Arabic gum as feed additives for tilapia and found they improved growth, survival and cellular responses to cold. Experts say the approach may help farms in cooler, subtropical areas but not very cold regions.
Alternative splicing linked to mammal lifespan
A study in Nature Communications compared alternative splicing across 26 mammal species (lifespans 2.2–37 years) and found splicing patterns better predict maximum lifespan than gene activity; the brain shows many lifespan-linked events controlled by RNA-binding proteins.