"The highlighted tokens are primarily Japanese content words—nouns, verbs, and adjectives—often marking key entities, actions, or attributes within phrases or sentences. These tokens tend to be semantically rich, frequently appearing at or near phrase boundaries, and are often accompanied by grammatical particles or function words that clarify their syntactic role. The pattern emphasizes the importance of content-bearing morphemes in conveying the main informational structure of Japanese text."
Score Type | Accuracy | Precision | Recall | F1 score | TPR | TNR | FPR | FNR |
---|---|---|---|---|---|---|---|---|
detection | 0.69 | 0.913 | 0.42 | 0.575 | 0.42 | 0.96 | 0.04 | 0.58 |
fuzz | 0.69 | 0.852 | 0.46 | 0.597 | 0.46 | 0.92 | 0.08 | 0.54 |