def run_geo(self):
train_ids, tune_ids, test_ids = self.get_folds()
train_nl = open('%s/train.nl' % self.config.experiment_dir, 'w')
train_nl_lm = open('%s/train.nl.lm' % self.config.experiment_dir, 'w')
train_nl_np = open('%s/train.np.nl' % self.config.experiment_dir, 'w')
train_mrl = open('%s/train.mrl' % self.config.experiment_dir, 'w')
train_mrl_lm = open('%s/train.mrl.lm' % self.config.experiment_dir, 'w')
train_mrl_np = open('%s/train.np.mrl' % self.config.experiment_dir, 'w')
train_fun = open('%s/train.fun' % self.config.experiment_dir, 'w')
unlabeled_nl = open('%s/unlabeled.nl' % self.config.experiment_dir, 'w')
tune_nl = open('%s/tune.nl' % self.config.experiment_dir, 'w')
tune_mrl = open('%s/tune.mrl' % self.config.experiment_dir, 'w')
test_nl = open('%s/test.nl' % self.config.experiment_dir, 'w')
test_mrl = open('%s/test.mrl' % self.config.experiment_dir, 'w')
test_fun = open('%s/test.fun' % self.config.experiment_dir, 'w')
corpus = ET.parse('%s/corpus-true.xml' % self.config.data_dir)
corpus_root = corpus.getroot()
counter = 0
#stop_labeling = False
for node in corpus_root.findall('example'):
nl = node.find("nl[@lang='%s']" % self.config.lang).text
nl = self.preprocess_nl(nl)
fun = node.find("mrl[@lang='geo-funql']").text
fun = self.preprocess_fun(fun)
#fun = self.replace_specials(fun)
mrl = util.fun_to_mrl(fun)
eid = int(node.attrib['id'])
unlabel_this = (counter >= 10 * self.config.lfrac)
counter += 1
counter %= 10
if eid in tune_ids:
print >>tune_nl, nl
print >>tune_mrl, mrl
elif eid in train_ids and not unlabel_this:
print >>train_nl, nl
print >>train_mrl, mrl
print >>train_fun, fun
print >>train_nl_np, nl
print >>train_mrl_np, mrl
print >>train_nl_lm, '<s>', nl, '</s>'
print >>train_mrl_lm, '<s>', mrl, '</s>'
elif eid in train_ids and unlabel_this:
print >>unlabeled_nl, nl
elif eid in test_ids:
print >>test_nl, nl
print >>test_mrl, mrl
print >>test_fun, fun
nplist = ET.parse('%s/nps-true.xml' % self.config.data_dir)
nplist_root = nplist.getroot()
for node in nplist_root.findall('example'):
fun = node.find("mrl[@lang='geo-funql']").text
fun = self.preprocess_fun(fun)
#fun = self.replace_specials(fun)
mrl = util.fun_to_mrl(fun)
big_np = len(mrl.split()) > 1
if (self.config.np_type == 'big' and not big_np) or \
(self.config.np_type == 'small' and big_np):
continue
for nl_node in node.findall("nl[@lang='%s']" % self.config.lang):
nl = nl_node.text
nl = self.preprocess_nl(nl)
for i in range(self.NP_WEIGHT):
print >>train_nl_np, nl
print >>train_mrl_np, mrl
print >>train_nl_lm, nl
print >>train_mrl_lm, mrl
train_nl.close()
train_nl_lm.close()
train_mrl.close()
train_mrl_lm.close()
train_fun.close()
test_nl.close()
test_mrl.close()
test_fun.close()
tune_nl.close()
tune_mrl.close()
def run_atis(self):
train_nl = open('%s/train.nl' % self.config.experiment_dir, 'w')
train_nl_lm = open('%s/train.nl.lm' % self.config.experiment_dir, 'w')
train_nl_np = open('%s/train.np.nl' % self.config.experiment_dir, 'w')
train_mrl = open('%s/train.mrl' % self.config.experiment_dir, 'w')
train_mrl_lm = open('%s/train.mrl.lm' % self.config.experiment_dir, 'w')
train_mrl_np = open('%s/train.np.mrl' % self.config.experiment_dir, 'w')
train_fun = open('%s/train.fun' % self.config.experiment_dir, 'w')
tune_nl = open('%s/tune.nl' % self.config.experiment_dir, 'w')
tune_mrl = open('%s/tune.mrl' % self.config.experiment_dir, 'w')
test_nl = open('%s/test.nl' % self.config.experiment_dir, 'w')
test_mrl = open('%s/test.mrl' % self.config.experiment_dir, 'w')
test_fun = open('%s/test.fun' % self.config.experiment_dir, 'w')
if self.config.run == 'debug':
with open('%s/atis-train.sem' % self.config.data_dir) as data_file:
counter = 0
for line in data_file:
nl, slot = line.split('<=>', 1)
nl = self.preprocess_nl(nl)
slot = self.replace_specials(slot)
fun = self.slot_to_fun(slot)
mrl = util.fun_to_mrl(fun, True)
if counter % 4 in (0,1):
print >>train_nl, nl
print >>train_mrl, mrl
print >>train_fun, fun
print >>train_nl_np, nl
print >>train_mrl_np, mrl
print >>train_nl_lm, '<s>', nl, '</s>'
print >>train_mrl_lm, '<s>', mrl, '</s>'
elif counter % 4 == 2:
print >>tune_nl, nl
print >>tune_mrl, mrl
else:
print >>test_nl, nl
print >>test_mrl, mrl
print >>test_fun, fun
counter += 1
else:
train_path = '%s/atis-train.sem' % self.config.data_dir
if self.config.run == 'dev':
tune_path = train_path
test_path = '%s/atis-dev.sem' % self.config.data_dir
elif self.config.run == 'test':
tune_path = '%s/atis-dev.sem' % self.config.data_dir
test_path = '%s/atis-test.sem' % self.config.data_dir
with open(train_path) as train_file:
for line in train_file:
nl, slot = line.split('<=>', 1)
nl = self.preprocess_nl(nl)
slot = self.replace_specials(slot)
fun = self.slot_to_fun(slot)
mrl = util.fun_to_mrl(fun, True)
print >>train_nl, nl
print >>train_mrl, mrl
print >>train_fun, fun
print >>train_nl_np, nl
print >>train_mrl_np, mrl
print >>train_nl_lm, '<s>', nl, '</s>'
print >>train_mrl_lm, '<s>', mrl, '</s>'
with open(tune_path) as tune_file:
for line in tune_file:
nl, slot = line.split('<=>', 1)
nl = self.preprocess_nl(nl)
slot = self.replace_specials(slot)
fun = self.slot_to_fun(slot)
mrl = util.fun_to_mrl(fun, True)
print >>tune_nl, nl
print >>tune_mrl, mrl
with open(test_path) as test_file:
for line in test_file:
nl, slot = line.split('<=>', 1)
nl = self.preprocess_nl(nl)
slot = self.replace_specials(slot)
fun = self.slot_to_fun(slot)
mrl = util.fun_to_mrl(fun, True)
print >>test_nl, nl
print >>test_mrl, mrl
print >>test_fun, fun
for np_name in os.listdir('%s/db' % self.config.data_dir):
np_path = '%s/db/%s' % (self.config.data_dir, np_name)
with open(np_path) as np_file:
for line in np_file:
names = re.findall(r'"([^"]+)"', line)
for name in names:
nl = name
mrl = "%s" % self.replace_specials(name)
mrl = mrl.replace(' ', '_')
mrl = mrl + '@s'
print >>train_nl_np, nl
print >>train_mrl_np, mrl
print >>train_nl_lm, nl
print >>train_mrl_lm, mrl
train_nl.close()
train_nl_lm.close()
train_mrl.close()
train_mrl_lm.close()
train_fun.close()
test_nl.close()
test_mrl.close()
test_fun.close()
tune_nl.close()
tune_mrl.close()
def run_robo(self):
train_ids, tune_ids, test_ids = self.get_folds()
tune_ids = test_ids
train_nl = open('%s/train.nl' % self.config.experiment_dir, 'w')
train_nl_lm = open('%s/train.nl.lm' % self.config.experiment_dir, 'w')
train_nl_np = open('%s/train.np.nl' % self.config.experiment_dir, 'w')
train_mrl = open('%s/train.mrl' % self.config.experiment_dir, 'w')
train_mrl_lm = open('%s/train.mrl.lm' % self.config.experiment_dir, 'w')
train_mrl_np = open('%s/train.np.mrl' % self.config.experiment_dir, 'w')
train_fun = open('%s/train.fun' % self.config.experiment_dir, 'w')
tune_nl = open('%s/tune.nl' % self.config.experiment_dir, 'w')
tune_mrl = open('%s/tune.mrl' % self.config.experiment_dir, 'w')
test_nl = open('%s/test.nl' % self.config.experiment_dir, 'w')
test_mrl = open('%s/test.mrl' % self.config.experiment_dir, 'w')
test_fun = open('%s/test.fun' % self.config.experiment_dir, 'w')
corpus = ET.parse('%s/corpus.xml' % self.config.data_dir)
corpus_root = corpus.getroot()
for node in corpus_root.findall('example'):
nl = node.find("nl[@lang='%s']" % self.config.lang).text
nl = self.preprocess_nl(nl)
clang = node.find("mrl[@lang='robocup-clang']").text
clang = self.replace_specials(clang)
fun = self.clang_to_fun(clang)
#print fun
mrl = util.fun_to_mrl(fun)
eid = int(node.attrib['id'])
if eid in tune_ids:
print >>tune_nl, nl
print >>tune_mrl, mrl
elif eid in train_ids:
print >>train_nl, nl
print >>train_mrl, mrl
print >>train_fun, fun
print >>train_nl_np, nl
print >>train_mrl_np, mrl
print >>train_nl_lm, '<s>', nl, '</s>'
print >>train_mrl_lm, '<s>', mrl, '</s>'
if eid in test_ids:
#elif eid in test_ids:
print >>test_nl, nl
print >>test_mrl, mrl
print >>test_fun, fun
nps_file = open('%s/names' % self.config.data_dir)
while True:
line = nps_file.readline()
if not line:
break
nl = nps_file.readline().strip()[3:]
nl = self.preprocess_nl(nl)
nps_file.readline()
nps_file.readline()
while True:
line = nps_file.readline().strip()
if line == '':
break
m = re.match('^\*n:(Num|Unum|Ident) -> \(\{ (\S+) \}\)$', line)
mrl = m.group(2) + '@0'
for i in range(self.NP_WEIGHT):
print >>train_nl_np, nl
print >>train_mrl_np, mrl
print >>train_nl_lm, nl
print >>train_mrl_lm, mrl
train_nl.close()
train_nl_lm.close()
train_mrl.close()
train_mrl_lm.close()
train_fun.close()
test_nl.close()
test_mrl.close()
test_fun.close()
tune_nl.close()
tune_mrl.close()
