Table 6 The detailed evaluation results with the state-of-the-art approach (CamemBERT) averaged among 10 runs and best results corresponding to the experiment with maximum weighted accuracy value.
From: A manually annotated corpus in French for the study of urbanization and the natural risk prevention
Class | Precision | Recall | F1 score | Accuracy (best) | Accuracy* (max) | |||
|---|---|---|---|---|---|---|---|---|
avg | best | avg | best | avg | best | |||
Verifiable | 0.82 | 0.83 | 0.85 | 0.86 | 0.84 | 0.84 | 0.90 | 0.89 |
Non-verifiable | 0.84 | 0.82 | 0.89 | 0.96 | 0.86 | 0.88 | ||
Informative | 0.83 | 0.85 | 0.83 | 0.85 | 0.83 | 0.85 | ||
Not pertinent | 0.95 | 0.98 | 0.91 | 0.89 | 0.93 | 0.93 | ||
