When a Nation Speaks: Machine Learning and NLP in People's Sentiment Analysis During Bangladesh's 2024 Mass Uprising

2025-12-17

AI Summary

This study addresses the scarcity of sentiment analysis research for Bengali in crisis contexts. Method: We conduct the first empirical analysis during the 2024 mass uprising in Bangladesh, constructing a manually annotated Bengali sentiment dataset comprising 2,028 news headlines, categorized into three emotion classes—anger, hope, and despair—to fill a critical gap in low-resource language sentiment analysis under civic unrest. We propose a context-aware topic–sentiment joint modeling framework that integrates LDA with a fine-tuned, Bengali-specific Transformer model, and comparatively evaluate it against SVM, logistic regression, mBERT, and XLM-RoBERTa. Contribution/Results: Our model achieves 73% accuracy—significantly outperforming mBERT (67%), XLM-RoBERTa (71%), and traditional classifiers (70%). Furthermore, the analysis uncovers dynamic public sentiment shifts triggered by crisis events such as internet shutdowns, offering novel insights into emotion evolution during political upheaval.

Technology Category

• Natural Language Processing: Sentiment Analysis, Stylistic Analysis, and Argument Mining
• Machine Learning: Evaluation and Analysis
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Web Mining and Content Analysis: Sentiment analysis and opinion mining
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

Sentiment analysis, an emerging research area within natural language processing (NLP), has primarily been explored in contexts like elections and social media trends, but there remains a significant gap in understanding emotional dynamics during civil unrest, particularly in the Bangla language. Our study pioneers sentiment analysis in Bangla during a national crisis by examining public emotions amid Bangladesh's 2024 mass uprising. We curated a unique dataset of 2,028 annotated news headlines from major Facebook news portals, classifying them into Outrage, Hope, and Despair. Through Latent Dirichlet Allocation (LDA), we identified prevalent themes like political corruption and public protests, and analyzed how events such as internet blackouts shaped sentiment patterns. It outperformed multilingual transformers (mBERT: 67%, XLM-RoBERTa: 71%) and traditional machine learning methods (SVM and Logistic Regression: both 70%). These results highlight the effectiveness of language-specific models and offer valuable insights into public sentiment during political turmoil.

Problem

• Analyzes Bangla sentiment during Bangladesh's 2024 mass uprising.
• Addresses the gap in understanding emotions during civil unrest.
• Evaluates language-specific models versus multilingual and traditional methods.

Innovation

• Bangla-specific sentiment analysis during national crisis
• LDA theme identification from annotated Facebook news headlines
• Language-specific models outperforming multilingual transformers and traditional ML