The paper outlines an approach for the ‘Arabic AI Tasks Evaluation (ArAiEval) Shared Task 2023’, focusing on detection of disinformation and persuasion techniques. Using transformer-based models fine-tuned on Arabic language, the authors aim to identify these elements in tweets and news articles. Despite challenges such as the evolving nature of disinformation and the high volume of data, the paper demonstrates success in detecting these techniques. The authors use ensemble methods to improve system performance, achieving high scores in the ArAiEval tasks.
Publication date: 1 Dec 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2311.18730
