From Scarcity to Capability: Empowering Fake News Detection in Low-Resource Languages with LLMs

TL;DR

The paper tackles fake news detection in low-resource Bangla by introducing BanFakeNews-2.0, a large, diverse dataset of 60k Bangla articles (47k authentic, 13k fake) across 13 categories, plus a 1,000-item independent test set. It benchmarks traditional linguistic features, transformer-based BERT models, and fine-tuned large language models using QLoRA, revealing that BLOOM-560M and m-BERT-unc offer strong performance, while dataset diversity enhances generalization. Key contributions include the expanded dataset, an independent test set, extensive baselines, and public release on GitHub to spur research and practical applications in Bangla fake news detection. The work demonstrates that diverse, well-annotated data paired with modern modeling approaches can significantly improve detection in low-resource languages, enabling more robust monitoring and mitigation of misinformation.

Abstract

The rapid spread of fake news presents a significant global challenge, particularly in low-resource languages like Bangla, which lack adequate datasets and detection tools. Although manual fact-checking is accurate, it is expensive and slow to prevent the dissemination of fake news. Addressing this gap, we introduce BanFakeNews-2.0, a robust dataset to enhance Bangla fake news detection. This version includes 11,700 additional, meticulously curated fake news articles validated from credible sources, creating a proportional dataset of 47,000 authentic and 13,000 fake news items across 13 categories. In addition, we created a manually curated independent test set of 460 fake and 540 authentic news items for rigorous evaluation. We invest efforts in collecting fake news from credible sources and manually verified while preserving the linguistic richness. We develop a benchmark system utilizing transformer-based architectures, including fine-tuned Bidirectional Encoder Representations from Transformers variants (F1-87\%) and Large Language Models with Quantized Low-Rank Approximation (F1-89\%), that significantly outperforms traditional methods. BanFakeNews-2.0 offers a valuable resource to advance research and application in fake news detection for low-resourced languages. We publicly release our dataset and model on Github to foster research in this direction.