AfrIFact: Cultural Information Retrieval, Evidence Extraction and Fact Checking for African Languages

2026-04-01

AI Summary

This study addresses the growing threat of online misinformation in low-resource African language communities, particularly concerning health and cultural topics, where automated fact-checking tools are scarce. To bridge this gap, the authors introduce AfrIFact, an end-to-end multilingual fact-checking dataset spanning ten African languages alongside English, establishing the first comprehensive benchmark for information retrieval, evidence extraction, and claim verification in these languages. The proposed approach leverages cross-lingual embeddings and AfriqueQwen-14B, a large language model tailored for African languages, enhanced through few-shot prompting and task-specific fine-tuning. Experiments demonstrate that few-shot prompting improves fact-checking accuracy by up to 43%, with further gains of 26% achieved through fine-tuning. The study also reveals that cultural and news-related documents are more readily retrievable than medical ones, highlighting limitations in current models’ cross-lingual retrieval and multilingual verification capabilities, thereby advancing research on trustworthy information access in low-resource linguistic contexts.

Abstract

Assessing the veracity of a claim made online is a complex and important task with real-world implications. When these claims are directed at communities with limited access to information and the content concerns issues such as healthcare and culture, the consequences intensify, especially in low-resource languages. In this work, we introduce AfrIFact, a dataset that covers the necessary steps for automatic fact-checking (i.e., information retrieval, evidence extraction, and fact checking), in ten African languages and English. Our evaluation results show that even the best embedding models lack cross-lingual retrieval capabilities, and that cultural and news documents are easier to retrieve than healthcare-domain documents, both in large corpora and in single documents. We show that LLMs lack robust multilingual fact-verification capabilities in African languages, while few-shot prompting improves performance by up to 43% in AfriqueQwen-14B, and task-specific fine-tuning further improves fact-checking accuracy by up to 26%. These findings, along with our release of the AfrIFact dataset, encourage work on low-resource information retrieval, evidence retrieval, and fact checking.

Problem

• fact-checking
• low-resource languages
• information retrieval
• evidence extraction
• African languages

Innovation

• low-resource languages
• cross-lingual retrieval
• fact-checking
• few-shot prompting
• multilingual LLMs