A part-of-speech tagger using averaged perceptrong and tagging history
You first should prepare 4 files:
- Preliminary segmented word corpus (train.words, dev.words)
- POS corpus corresponding above words (train.pos, dev.pos)
To train the tagging model using:
-
Word unigram
-
POS bigram
-
Word window with width = 3
-
5 POS histories,
-
10 training epoch run below:
postagger-train.py train dev model 1 2 3 5 10
- Yusuke Oda (@odashi)
We are counting more contributions from you.
If you find an issue, please contact Y.Oda
- yus.takara (at) gmail.com
- @odashi_t on Twitter