Table 3. Accuracy on Chinese linguistic phenomena.
The values with the bold style are the maximum values in each column among these methods or models.
| Datasets | Model | baˇ(“把”) | bèi(“被”) | -zhe(“着”) | -le(“了”) | -guo(“过”) |
|---|---|---|---|---|---|---|
| UD | random BERT | 11.65 | 8.06 | 6.45 | 7.62 | 10.00 |
| positional offset | 38.83 | 71.79 | 100.00 | 89.24 | 100.00 | |
| Chinses BERT |
56.31 (6–10) |
87.52 (6–11) |
100.00 (2–1) |
89.97 (4–8) |
100.00 (2–1) |
|
| CDT | random BERT | 6.51 | 7.22 | 5.05 | 5.55 | 4.38 |
| positional offset | 31.64 | 59.18 | 99.03 | 90.40 | 98.21 | |
| Chinses BERT |
84.66 (4–1) |
90.25 (6–11) |
99.03 (3–9) |
90.65 (4–8) |
98.21 (3–9) |