چکیده مقاله
The rapid spread of disinformation across digital channels threatens public discourse, the integrity of democratic processes, and the resilience of global health infrastructures Manual fact checking simply cannot keep pace with the torrents of content being produced, creating urgent demand for scalable, automated approaches to disinformation detection This survey presents a thorough evaluation of the latest Machine Learning ML and Deep Learning DL techniques devised to counter fake news, spotlighting transformer based models, hybrid architectures, and graph augmented networks We contrast conventional ML classifiers including Support Vector Machines and Random Forests against DL counterparts such as convolutional networks, LSTMs, and state of the art attention architectures like BERT, ROBERTa, and GAT BERT The review also catalogs emerging benchmark datasets ReCOVery, COVID FND, WELFake , introduces innovative evaluation metrics, and highlights promising research trajectories real time identification, multimodal integration, multilingual scaling, and Explainable AI XAI By distilling findings from the current body of literature, this manuscript seeks to inform the design of forthcoming detection systems that excel in accuracy while remaining interpretable, resilient, and aligned with ethical deployment in society
کلیدواژهها
نویسندگان
Arian Hajati
Department of Computer, CT C , Islamic Azad University, Tehran, Iran
Azita Shirazipour
Department of Computer, CT C , Islamic Azad University, Tehran, Iran
Seyed Javad Mirabedini
Department of Computer, CT C , Islamic Azad University, Tehran, Iran
شیوه ارجاع
Hajati, Arian and Shirazipour, Azita and Mirabedini, Seyed Javad,1404,Machine Learning and Deep Learning Approaches for Fake News Detection: A Comprehensive Survey,The Second National Conference on Data Science in Engineering Applications,Tabriz
ارائهشده در
مجموعه مقالات دومین کنفرانس ملی علم داده در کاربردهای مهندسی17 مهر 1404 · تبریز