A federated deep learning framework with distributed hybrid character-level and attention mechanisms for scalable and cost-efficient fake news detection.

As shown in Table 13 , the comparison between the proposed COASBC and Federated CharBERT COASBC models yields p-values less than 0.001 for both the LIAR (Augmented), ISOT datasets and FakeNewsNet, and WELFake Dataset, indicating highly significant improvements.

also in 132,142 other papers