Research Article
Multilevel Feature Fusion-Based GCN for Rumor Detection with Topic Relevance Mining
Table 1
Distribution of annotations of rumors and nonrumors for the five breaking news in the dataset.
| Event | Rumors | Nonrumours | Total |
| Charlie Hebdo | 458 (22.0%) | 1,621 (78.0%) | 2,079 | Ferguson | 284 (24.8%) | 859 (75.2%) | 1,143 | Germanwings crash | 238 (50.7%) | 231 (49.3%) | 469 | Ottawa shooting | 470 (52.8%) | 420 (47.2%) | 890 | Sydney siege | 522 (42.8%) | 699 (57.2%) | 1,221 | Total | 1,972 (34.0%) | 3,830 (66.0%) | 5,802 |
|
|