"The highlighted tokens frequently correspond to named entities, locations, organizations, and key connecting words in multiple languages, often marking the start or continuation of proper nouns, place names, or important phrases within travel routes, political contexts, or descriptive sequences. These tokens are often found at the boundaries of multi-word expressions or in the context of enumerating places, people, or organizations, and they include both content words and function words that are crucial for linking or specifying entities in multilingual text."
Score Type | Accuracy | Precision | Recall | F1 score | TPR | TNR | FPR | FNR |
---|---|---|---|---|---|---|---|---|
detection | 0.48 | 0.482 | 0.54 | 0.509 | 0.54 | 0.42 | 0.58 | 0.46 |
fuzz | 0.41 | 0.435 | 0.6 | 0.504 | 0.6 | 0.22 | 0.78 | 0.4 |