UNRESTSENT200K Dataset for Bangla Crisis Sentiment Analysis
September 16, 2026
The UNRESTSENT200K dataset provides 200,000 annotated Bangla Facebook and YouTube comments from the 2024 Bangladesh uprising. Benchmarking shows that providing parent-post context significantly improves performance for fine-tuned encoders and LLMs in low-resource crisis scenarios.
HOW THIS AFFECTS YOU
●
researcherYou can use this dataset to benchmark LLM performance in low-resource, high-volatility linguistic contexts.