def train():
MODE = 'train'
featExt = extractors.get(FLAGS.feature_extarctor)
sents = io.transform_conll_sents(FLAGS.train_data, FLAGS.only_projective,
FLAGS.unlex)
trainer = MLTrainerActionDecider(ml.MultitronParameters(3),
ArcStandardParsingOracle(), featExt)
p = ArcStandardParser2(trainer)
total = len(sents)
random.seed("seed")
for x in xrange(FLAGS.epoch):
random.shuffle(sents)
logging.info("iter %s/%s", x + 1, FLAGS.epoch)
logging.info(" shuffle data ...")
for i, sent in enumerate(sents):
if i % 500 == 0:
logging.info(" step %s/%s ...", i, total)
try:
d = p.parse(sent)
except Exception as e:
logging.info("prob in sent: %s", i)
logging.info("\n".join([
"%s %s %s %s" % (t['id'], t['form'], t['tag'], t['parent'])
for t in sent
]))
raise e
with open(FLAGS.model, "w") as fout:
logging.info("save model file to disk [%s] ...", FLAGS.model)
trainer.save(fout)
def train():
'''
Train Model
'''
MODE = 'train'
TRAIN_OUT_FILE = FLAGS.model
if FLAGS.externaltrainfile:
'''
create feature vector files for training with an external classifier. If you don't know what it means,
just ignore this option. The model file format is the same as Megam's.
'''
MODE = 'write'
TRAIN_OUT_FILE = FLAGS.externaltrainfile
featExt = extractors.get(FLAGS.feature_extarctor)
sents = io.transform_conll_sents(FLAGS.train_data, FLAGS.only_projective,
FLAGS.unlex)
if MODE == "write":
fout = file(TRAIN_OUT_FILE, "w")
trainer = LoggingActionDecider(
ArcEagerParsingOracle(pop_when_can=FLAGS.lazypop), featExt, fout)
p = ArcEagerParser(trainer)
for i, sent in enumerate(sents):
sys.stderr.write(". %s " % i)
sys.stderr.flush()
d = p.parse(sent)
sys.exit()
if MODE == "train":
fout = file(TRAIN_OUT_FILE, "w")
nactions = 4
trainer = MLTrainerActionDecider(
ml.MultitronParameters(nactions),
ArcEagerParsingOracle(pop_when_can=FLAGS.lazypop), featExt)
p = ArcEagerParser(trainer)
import random
random.seed("seed")
total = len(sents)
for x in xrange(FLAGS.epoch): # epoch
logging.info("iter %s/%s", x + 1, FLAGS.epoch)
logging.info(" shuffle data ...")
random.shuffle(sents)
for i, sent in enumerate(sents):
if i % 500 == 0:
logging.info(" step %s/%s ...", i, total)
try:
d = p.parse(sent)
except IndexError as e:
logging.info("prob in sent: %s", i)
logging.info("\n".join([
"%s %s %s %s" %
(t['id'], t['form'], t['tag'], t['parent'])
for t in sent
]))
raise e
logging.info("save model file to disk [%s] ...", TRAIN_OUT_FILE)
trainer.save(fout)
def test():
featExt = extractors.get(FLAGS.feature_extarctor)
sents = io.transform_conll_sents(FLAGS.test_data, FLAGS.only_projective,
FLAGS.unlex)
p = ArcStandardParser2(
MLActionDecider(ml.MulticlassModel(FLAGS.model, True), featExt))
good = 0.0
bad = 0.0
complete = 0.0
with open(FLAGS.test_results, "w") as fout:
for i, sent in enumerate(sents):
mistake = False
sgood = 0.0
sbad = 0.0
fout.write("%s %s %s\n" % ("@@@", i, good / (good + bad + 1)))
try:
d = p.parse(sent)
except MLTrainerWrongActionException:
continue
sent = d.annotate(sent)
for tok in sent:
# print tok['id'], tok['form'], "_",tok['tag'],tok['tag'],"_",tok['pparent'],"_ _ _"
if FLAGS.ignore_punc and tok['form'][0] in "`',.-;:!?{}":
continue
if tok['parent'] == tok['pparent']:
good += 1
sgood += 1
else:
bad += 1
sbad += 1
mistake = True
if not mistake: complete += 1
fout.write("%s\n" % (sgood / (sgood + sbad)))
print("accuracy:", good / (good + bad))
print("complete:", complete / len(sents))
def test():
'''
Test Model
'''
logging.info("test ...")
featExt = extractors.get(FLAGS.feature_extarctor)
p = ArcEagerParser(
MLActionDecider(ml.MulticlassModel(FLAGS.model), featExt))
good = 0.0
bad = 0.0
complete = 0.0
# main test loop
reals = set()
preds = set()
with open(FLAGS.test_results, "w") as fout:
sents = io.transform_conll_sents(FLAGS.test_data,
FLAGS.only_projective, FLAGS.unlex)
for i, sent in enumerate(sents):
sgood = 0.0
sbad = 0.0
mistake = False
sys.stderr.write("%s %s %s\n" % ("@@@", i, good /
(good + bad + 1)))
try:
d = p.parse(sent)
except MLTrainerWrongActionException:
# this happens only in "early update" parsers, and then we just go on to
# the next sentence..
continue
sent = d.annotate_allow_none(sent)
for tok in sent:
if FLAGS.ignore_punc and tok['form'][0] in "`',.-;:!?{}":
continue
reals.add((i, tok['parent'], tok['id']))
preds.add((i, tok['pparent'], tok['id']))
if tok['pparent'] == -1:
continue
if tok['parent'] == tok['pparent'] or tok['pparent'] == -1:
good += 1
sgood += 1
else:
bad += 1
sbad += 1
mistake = True
if FLAGS.unlex:
io.out_conll(sent, parent='pparent', form='oform')
else:
io.out_conll(sent, parent='pparent', form='form')
if not mistake:
complete += 1
# sys.exit()
logging.info("test result: sgood[%s], sbad[%s]", sgood, sbad)
if sgood > 0.0 and sbad > 0.0:
fout.write("%s\n" % (sgood / (sgood + sbad)))
logging.info("accuracy: %s", good / (good + bad))
logging.info("complete: %s", complete / len(sents))
preds = set([(i, p, c) for i, p, c in preds if p != -1])
logging.info("recall: %s",
len(preds.intersection(reals)) / float(len(reals)))
logging.info("precision: %s",
len(preds.intersection(reals)) / float(len(preds)))
logging.info("assigned: %s", len(preds) / float(len(reals)))
gflags.DEFINE_boolean("ignore_punc", True, "Ignore punctuation in evaluation.")
gflags.DEFINE_boolean("only_proj", True,
"If true, prune non-projective sentences in training.")
gflags.DEFINE_boolean("add_dep_label", True,
"If true, replace the '_' label with 'dep'.")
gflags.DEFINE_integer("random_seed", 0, "Random seed.")
gflags.DEFINE_integer("save_every", 0, "Dump a model every k iterations.")
args = FLAGS(sys.argv)
print args
DATA_FILE = args[1]
featExt = extractors.get(FLAGS.feature_extractor)
sents = list(io.conll_to_sents(file(DATA_FILE)))
if FLAGS.train and (True or FLAGS.only_proj):
import isprojective
sents = [s for s in sents if isprojective.is_projective(s)]
if FLAGS.add_dep_label:
for sent in sents:
for tok in sent:
if tok['prel'] == '_': tok['prel'] = "dep"
EXPLORE = 1
LABELED = True