* Add script to train the dictionary-supervised supersense tagger

This commit is contained in:
Matthew Honnibal 2015-07-06 09:06:22 +02:00
parent 88a4e53fcb
commit 2e4cfe5255

View File

@ -0,0 +1,73 @@
#!/usr/bin/env python
from __future__ import division
from __future__ import unicode_literals
import os
from os import path
import shutil
import codecs
import random
import plac
import cProfile
import pstats
import re
from spacy.en import English
def score_model(nlp, semcor_docs):
n_right = 0
n_wrong = 0
n_multi = 0
for dnum, paras in semcor_docs:
for pnum, para in paras:
for snum, sent in para:
words = [t.orth for t in sent]
tokens = nlp.tokenizer.tokens_from_list(words)
nlp.tagger(tokens)
nlp.senser(tokens)
for i, token in enumerate(tokens):
if '_' in sent[i].orth:
n_multi += 1
elif sent[i].supersense != 'NO_SENSE':
n_right += token.sense_ == sent[i].supersense
n_wrong += token.sense_ != sent[i].supersense
return n_multi, n_right, n_wrong
def train(Language, model_dir, docs, annotations, report_every=1000, n_docs=1000):
wsd_model_dir = path.join(model_dir, 'wsd')
if path.exists(pos_model_dir):
shutil.rmtree(pos_model_dir)
os.mkdir(wsd_model_dir)
Config.write(wsd_model_dir, 'config', seed=seed)
nlp = Language(data_dir=model_dir)
for doc in corpus:
tokens = nlp(doc, senser=False)
loss += nlp.senser.train(tokens)
if i and not i % report_every:
acc = score_model(nlp, dev_docs)
print loss, n_right / (n_right + n_wrong)
nlp.senser.end_training()
nlp.vocab.strings.dump(path.join(model_dir, 'vocab', 'strings.txt'))
@plac.annotations(
docs_db_loc=("Location of the documents SQLite database"),
model_dir=("Location of the models directory"),
n_docs=("Number of training documents", "option", "n", int),
verbose=("Verbose error reporting", "flag", "v", bool),
debug=("Debug mode", "flag", "d", bool),
)
def main(train_loc, dev_loc, model_dir, n_docs=0):
train_docs = DocsDB(train_loc)
dev_docs = read_semcor(dev_loc)
train(English, model_dir, train_docs, dev_docs, report_every=10, n_docs=1000):
if __name__ == '__main__':
plac.call(main)