| Abstrakt: |
The widespread dissemination of fake news through media platforms has driven the research community to focus on developing methods for its detection. Researchers have made numerous efforts to build effective models that can automatically identify fake news articles, leveraging advances in natural language processing and machine learning. Approaches using linguistic features and neural networks have shown promise, yet the performance of these methods heavily depends on the availability of large, annotated datasets. While several benchmark datasets for fake news detection exist, particularly for English, there is a notable scarcity of resources for low-resource languages such as Arabic. This study addresses this gap by exploring datasets for fake news detection in Arabic. Our study extends this by surveying a more comprehensive collection of 29 Arabic datasets. It provides an in-depth analysis of available datasets, evaluates their characteristics, and highlights the challenges of fake news detection in Arabic, such as the lack of multimedia data, limited diversity in news domains, insufficient dataset sizes, and the need for comprehensive benchmarks. This research contributes to the ongoing efforts to improve fake news detection in low-resource languages through advanced machine learning and deep learning techniques. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink