NLP-based approach to multilingual fake news detection through social media in low-resource languages: A review

Rapid dissemination of fake news via social media threatens society's stability, especially in multilingual and low-resource languages. This paper reviews the state-of-the-art NLP techniques applied in fake news detection across multiple languages, with a particular focus on low-resource languages. It discusses the main methods of text classification, sentiment analysis, and NER and their applications, advantages, and challenges in such settings. This paper also investigates how far techniques such as transfer learning, multilingual embeddings, and cross-lingual models can go toward meeting the challenges of linguistic diversity and scarcity of annotated data. Lastly, it identifies where the current state of the art fails and the implications of new directions for future improvement in model accuracy and adaptability for multilingual fake news detection. The findings of this paper not only enlighten the different challenges faced and innovations in this crucial area but also contribute to enabling the development of robust and inclusive solutions to mitigate the adverse effects of misinformation around the world.