"The highlighted tokens are primarily Hindi or Devanagari script morphemes, syllables, or short word fragments, often marking the start, end, or important part of proper nouns, place names, or key content words. There is a strong emphasis on tokens that function as grammatical markers, connectors, or are part of compound words, especially in names, locations, and official titles. The pattern reflects the segmentation of Hindi text into meaningful units, with frequent focus on morphemes that contribute to the structure and meaning of complex words or phrases."
Score Type | Accuracy | Precision | Recall | F1 score | TPR | TNR | FPR | FNR |
---|---|---|---|---|---|---|---|---|
detection | 0.91 | 1.0 | 0.82 | 0.901 | 0.82 | 1.0 | 0.0 | 0.18 |
fuzz | 0.89 | 0.933 | 0.84 | 0.884 | 0.84 | 0.94 | 0.06 | 0.16 |