๐ถ Brill tagger
= a transformation based POS-tagger
How
For each word in corpus:
- Tag the word with the most likely POS
- Choose a transformation that deterministically replaces an existing tag with a new one to improve the error rate
- Apply the transformation
Example transformations
# | From | To | Condition | Example |
---|---|---|---|---|
1 | NN | VB | Previous tag is TO | to/TO race/NN โ VB |
2 | VBP | VB | One of the previous tags is MD | might/MD vanish/VBP โ VB |