Large Arabic fact checking dataset created using AI language models
ARAFA: An LLM-Generated Arabic Fact-Checking Dataset
Computation and LanguageInformation Retrieval
Summary
Fact-checking Arabic texts automatically is hard because there aren’t many good datasets available. The authors created a large dataset called Arafa by using AI language models to make true and false claims from Arabic Wikipedia articles, along with evidence that supports or refutes these claims. The dataset includes nearly 182,000 pairs of claims and evidence labeled as true, false, or lacking enough information. Human reviewers confirmed the dataset quality, and models trained on it performed well at checking Arabic facts. This approach can also help build similar datasets for other languages with few resources.
What this means in practice
- •For arabic nlp engineers: Train and evaluate automatic fact-checking systems for Arabic text using a large, high-quality labeled dataset.
- •For machine learning platform developers: Create automated pipelines for generating and validating fact-checking datasets in low-resource languages based on this scalable framework.
Authors
Christophe Khalil, Shady Elbassuoni, Rida Assaf
Abstract
Automatic fact-checking poses a significant challenge in Arabic natural language processing due to the scarcity of datasets and resources. In this manuscript, we introduce Arafa, a new large-scale dataset for fact-checking in Modern Standard Arabic, constructed through an automated framework leveraging large language models (LLMs). The dataset was constructed through a three-step pipeline: (1) claim generation from Arabic Wikipedia pages with supporting textual evidence, (2) claim mutation to generate challenging counterfactual claims with refuting evidence, and (3) an automatic validation step to validate that the generated claims are either supported or refuted by their accompanying evidence, or if the evidence does not provide enough information to judge the validity of the claims. The resulting dataset comprises 181,976 claim-evidence pairs labeled as supported, refuted, or not enough information. Human evaluation carried out on a test sample from the dataset demonstrated strong inter-annotator agreement (kappa = 0.89) using Cohen's Kappa for supported claims and (kappa = 0.94) for refuted claims. Automatic validation based on a human-evaluated sample achieved 86% accuracy for supported claims and 88% for refuted ones. To showcase Arafa's value as a resource for automatic Arabic fact-checking, four open-source transformer-based models were fine-tuned using Arafa, with the top-performing model achieving a Macro F1-score of 77% on the test data. In addition to Arafa being the first large-scale dataset for Arabic fact-checking, our framework presents a scalable approach for developing similar resources for other low-resource languages.