"The highlighted tokens are primarily Korean morphemes, syllables, or words, often marking grammatical functions, noun or verb stems, and endings. There is frequent activation on common content and function morphemes, as well as on foreign or borrowed terms (such as Hangeul-related names). Additionally, corrupted or unreadable characters are sometimes marked, likely due to encoding issues. The pattern reflects a focus on meaningful linguistic units in Korean text, especially those that contribute to sentence structure or key content."
Score Type | Accuracy | Precision | Recall | F1 score | TPR | TNR | FPR | FNR |
---|---|---|---|---|---|---|---|---|
detection | 0.87 | 1.0 | 0.74 | 0.851 | 0.74 | 1.0 | 0.0 | 0.26 |
fuzz | 0.86 | 0.974 | 0.74 | 0.841 | 0.74 | 0.98 | 0.02 | 0.26 |