كنفرانس بين المللی مهندسی صنايع و سيستم ها

صفحه اصلی / یازدهمین كنفرانس بين المللی مهندسی صنايع و سيستم ها

Explainable Artificial Intelligence Approaches for Fake News Detection: A Comprehensive Study

نویسندگان :

Amirhossein Tohidi aghdam¹ Azita Shirazipour² Javad Mirabedini³

1- دانشگاه آزاد تهران مرکز 2- دانشگاه آزاد تهران مرکز 3- دانشگاه آزاد تهران مرکز

کلمات کلیدی :

Fake News Detection،Explainable Artificial Intelligence،Deep Learning،Graph Neural Networks،Multimodal Information Fusion،LIME-based explanations

چکیده :

The explosion of fake news in today’s digital landscape threatens the reliability of information on which societies depend, affecting everything from voting behavior to public trust. While existing machine learning and deep learning strategies for fake news detection (FND) have delivered encouraging accuracy, their opaque decision-making processes hinder the transparency and reliability required in high-stakes environments. This study surveys how Explainable Artificial Intelligence (XAI) approaches can strengthen FND, extending the inquiry to diverse data modalities and evolving network architectures. By integrating findings emerging since 2023, we chart the state of the art, spotlighting widely adopted XAI techniques—LIME, SHAP, and evolving attention mechanisms—alongside the latest transformer and graph- neural-net architectures. We also underscore the value of multimodal data fusion, where textual, visual, and social signals collaborate to raise both detection precision and the comprehensibility of the outcomes. The review candidly addresses persistent hurdles: balancing the demand for interpretable outputs and high accuracy, managing increased computational costs, and the absence of universally accepted benchmarks for evaluating explanations. Lastly, we propose priority avenues for future inquiry, aiming to produce trustworthy, transparent FND systems that can be accepted by both users and regulatory bodies.