Table 14 shows that the distribution of polysemous items predicted by Equation (4) is more similar to the distribution obtained with the best machine learning classifier (ML) than to the distribution of polysemous items in the gold standard (GS) for the QE cases. The distribution is estimated from the frequency over the 210 adjectives in the gold standard, and shown as absolute numbers.
Distribution of polysemous items and absolute numbers, according to the prediction (Equation (4); first column), in the machine learning (ML) results shown in Table 12 (second column), and in the gold standard (GS; third column).
. | Predicted . | ML . | GS . |
|---|---|---|---|
| QR | 15 | 22 | 23 |
| QE | 19 | 19 | 7 |
| ER | 5 | 9 | 6 |
. | Predicted . | ML . | GS . |
|---|---|---|---|
| QR | 15 | 22 | 23 |
| QE | 19 | 19 | 7 |
| ER | 5 | 9 | 6 |