"The highlighted tokens are predominantly morphemes, syllables, or name fragments within proper nouns, place names, and personal names, often from Turkish but also from other languages. These tokens frequently appear in transliterations, compound words, or as parts of culturally specific terms, indicating a focus on subword units that contribute to the identification and construction of named entities and culturally significant vocabulary."
Score Type | Accuracy | Precision | Recall | F1 score | TPR | TNR | FPR | FNR |
---|---|---|---|---|---|---|---|---|
detection | 0.87 | 0.849 | 0.9 | 0.874 | 0.9 | 0.84 | 0.16 | 0.1 |
fuzz | 0.66 | 0.6 | 0.96 | 0.738 | 0.96 | 0.36 | 0.64 | 0.04 |