Table 4.
Table 4 Partial Confusion Matrix Showing Distribution of Most Frequent POS Tagging Errors by ME Tagger Trained with WSJ and DSC (Data Set 3)
| Tagging Error | Reference Standard | ME Tagger trained with WSJ + DSC | % of Total | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| NN | JJ | NNP | VBD | VBN | CD | NNS | RB | VB | DT | |||
| JJ | 19% (49) | — | 2% (4) | 0% (1) | 1% (2) | — | — | — | 1% (3) | — | 24% (64) | |
| LS | 6% (17) | 0% (1) | 8% (20) | — | — | 3% (7) | — | — | — | 1% (2) | 19% (49) | |
| NN | — | 11% (28) | 1% (3) | 0% (1) | — | — | — | — | — | — | 13% (34) | |
| VBN | — | — | — | 8% (21) | — | — | — | — | — | — | 8% (22) | |
| VBD | — | — | — | — | 8% (20) | — | — | — | — | — | 8% (20) | |
| NNP | 3% (7) | 3% (9) | — | — | — | — | 0% (1) | — | — | — | 7% (18) | |
| NNS | 5% (13) | 0% (1) | — | — | — | — | — | — | — | — | 6% (15) | |
| CC | — | — | — | — | — | — | — | 1% (3) | — | — | 2% (6) | |
| IN | — | — | — | — | — | — | — | 1% (2) | — | — | 2% (5) | |
| RB | — | 1% (3) | — | — | — | — | — | — | — | — | 2% (5) | |
| % of Total | 36% (94) | 18% (47) | 11% (28) | 9% (23) | 8% (22) | 3% (8) | 3% (7) | 2% (6) | 2% (5) | 0% (3) | 90% (238) | |
JJ = adjective; LS = list item marker; NN = noun, singular or mass; VBN = verb, past participle; VBD = verb, past tense; NNP = noun, proper, singular; NNS = noun, plural; CC = conjunction, coordinating; IN = preposition or conjunction, subordinating; RB = adverb; CD = numeral, cardinal; VB = verb, base form; DT = determiner.
Number of errors shown in percentage of total errors, with counts in parentheses, for most frequent errors.