Table 9 5-Fold Cross-Validation results on FakeNewsNet dataset (Mean ± Standard Deviation).

From: Graph-augmented transformer ensemble framework for robust and scalable fake news detection in social media ecosystems

Configuration

Accuracy (%)

F1-score (%)

ROC-AUC (%)

GNN Only

90.1 ± 0.6

89.8 ± 0.7

91.0 ± 0.4

BERT Only

91.8 ± 0.5

91.6 ± 0.5

92.9 ± 0.3

RoBERTa Only

92.8 ± 0.5

92.5 ± 0.5

93.9 ± 0.3

BERT + GNN (no ensemble)

94.2 ± 0.4

94.0 ± 0.4

95.0 ± 0.3

Full Model (GETE Ensemble)

96.0 ± 0.3

95.8 ± 0.4

97.0 ± 0.2

Back to article page