Research Article
Rumor Detection Based on Knowledge Enhancement and Graph Attention Network
Table 4
Comparison of detection accuracy on Twitter.
| Twitter | Method accuracy | Rumors | Nonrumors | | Precision | F1 | Precision | F1 |
| Baseline | 0.548 | 0.581 | 0.558 | 0.577 | 0.548 | CNN | 0.653 | 0.655 | 0.643 | 0.663 | 0.653 | LSTM | 0.796 | 0.706 | 0.723 | 0.724 | 0.697 | RNN | 0.785 | 0.712 | 0.692 | 0.701 | 0.679 | GRU | 0.736 | 0.729 | 0.734 | 0.745 | 0.729 | BERT | 0.752 | 0.792 | 0.819 | 0.782 | 0.733 | Bi-GCN | 0.880 | 0.924 | 0.930 | 0.853 | 0.847 | Ours | 0.885 | 0.937 | 0.942 | 0.856 | 0.852 |
|
|