Automatic Fact-checking in English and Telugu

2025-09-30

AI Summary

Addressing cross-lingual misinformation propagation, particularly in low-resource languages. Method: We propose the first English–Telugu bilingual fact-checking framework, comprising: (1) construction of the first high-quality English–Telugu bilingual fact-checking dataset; (2) systematic evaluation of multilingual large language models (LLMs) under zero-shot and few-shot settings for veracity classification and explanation generation; and (3) investigation of cross-lingual transfer mechanisms. Contribution/Results: This work establishes the first Telugu fact-checking benchmark, empirically validating both the effectiveness and cross-lingual transferability of LLMs in low-resource bilingual settings. It significantly reduces reliance on manual verification. Experiments demonstrate state-of-the-art automatic fact-checking performance on both English and Telugu, providing a reproducible benchmark and methodological foundation for multilingual fact-checking research.

Technology Category

Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)Machine Learning: Large Multimodal Models (LMMs)Knowledge Representation and Reasoning: Knowledge Representation Languages

Application Category

Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web searchWeb Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web dataUser Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

False information poses a significant global challenge, and manually verifying claims is a time-consuming and resource-intensive process. In this research paper, we experiment with different approaches to investigate the effectiveness of large language models (LLMs) in classifying factual claims by their veracity and generating justifications in English and Telugu. The key contributions of this work include the creation of a bilingual English-Telugu dataset and the benchmarking of different veracity classification approaches based on LLMs.

Problem

• Automating fact-checking of claims in English and Telugu
• Evaluating LLMs for veracity classification and justification generation
• Creating a bilingual dataset and benchmarking classification methods

Innovation

• Created bilingual English-Telugu fact-checking dataset
• Benchmarked LLM approaches for veracity classification
• Generated justifications in both English and Telugu