def qc_script():
trainbase = '../data/QC/TREC/trimengqctrain'
translate_base = '../data/QC/translate/final_moses_eng2ch_train'
testbase = '../data/QC/Chinese_qc/finaltest'
validbase = '../data/QC/Chinese_qc/validset'
chtrainbase = '../data/QC/Chinese_qc/finaltrain.new'
# data quantity
logprefix = '../exp/qc_data_quantity'
accs = []
open(logprefix, 'w').close()
for i in xrange(1, 11):
cdep = False
dep = True
phr = True
lex = True
config_ = config(trainbase, 'eng', testbase, validbase, lex, phr, cdep, logprefix)
word2vec.rundown_config(config_)
process_qc.datasetConstructRundown_config(config_, dep, eng_portion=i)
# general running
batch_size = 120
feature_map = 110
filter_hs = [3]
accs.append(cnn_model.general_rundown(batch_size, feature_map, filter_hs, config_.lexicon))
with open(logprefix, 'w') as writer:
for acc in accs:
writer.write(str(acc) + '\t')
""" running example
def pr_script():
trainbase = '../data/Semantic/productReview/train.dat'
translate_base = '../data/Semantic/productReview/moses_train.dat'
testbase = '../data/Semantic/productReview/test.dat.new'
validbase = '../data/Semantic/productReview/validset'
# data quantity
logprefix = '../exp/pr_data_quantity'
accs = []
open(logprefix, 'w').close()
for i in xrange(1, 11):
cdep = False
dep = True
phr = True
lex = True
config_ = config(trainbase, 'eng', testbase, validbase, lex, phr, cdep, logprefix)
word2vec.rundown_config(config_)
process_qc.datasetConstructRundown_config(config_, dep, eng_portion=i)
# general running
batch_size = 180
feature_map = 50
filter_hs = [3]
accs.append(cnn_model.general_rundown(batch_size, feature_map, filter_hs, config_.lexicon))
with open(logprefix, 'w') as writer:
for acc in accs:
writer.write(str(acc) + '\t')
"""
def event_script_mono():
trainbase = '../data/Event/English/sub_train.dat'
translate_base = '../data/Event/translate/moses_train.dat'
chtrainbase = '../data/Event/Chinese/rest_test.dat'
testbase = '../data/Event/Chinese/sub_test.dat'
validbase = '../data/Event/Chinese/validset'
# data quantity
logprefix = '../exp/event_data_quantity'
accs = []
open(logprefix, 'w').close()
for i in xrange(1, 11):
cdep = False
dep = True
phr = True
lex = True
config_ = config(trainbase, 'eng', testbase, validbase, lex, phr, cdep, logprefix)
word2vec.rundown_config(config_)
process_qc.datasetConstructRundown_config(config_, dep, eng_portion=i)
# general running
batch_size = 160
feature_map = 130
filter_hs = [3]
accs.append(cnn_model.general_rundown(batch_size, feature_map, filter_hs, config_.lexicon))
with open(logprefix, 'w') as writer:
for acc in accs:
writer.write(str(acc) + '\t')
"""
def event_script2():
trainbase = '../data/Event/English/sub_train.dat'
translate_base = '../data/Event/translate/moses_train.dat'
chtrainbase = '../data/Event/Chinese/rest_test.dat.new'
testbase = '../data/Event/Chinese/sub_test.dat'
validbase = '../data/Event/Chinese/validset'
cdep = False
dep = True
phr = False
lex = False
hasmlphidden = False
logprefix = '../exp/event_ch'
filter_hss = [[3], [3, 4], [3, 4, 5]]
config_ = config(chtrainbase, 'ch', testbase, validbase, lex, phr, cdep, logprefix, usefscore=True)
word2vec.rundown_config(config_)
process_qc.datasetConstructRundown_config(config_, dep)
# cnn_model.structure_script(config_, filter_hss, hasmlphidden)
# general running english
batch_size = 170
feature_map = 90
filter_hs = [3, 4, 5]
cnn_model.general_rundown(batch_size, feature_map, filter_hs, config_)
def qc_script2():
trainbase = '../data/QC/TREC/trimengqctrain'
translate_base = '../data/QC/translate/final_moses_eng2ch_train'
testbase = '../data/QC/Chinese_qc/finaltest'
validbase = '../data/QC/Chinese_qc/validset'
chtrainbase = '../data/QC/Chinese_qc/finaltrain.new'
cdep = False
dep = True
phr = True
lex = True
hasmlphidden = False
logprefix = '../exp/qc_f_lex_phr'
filter_hss = [[1, 3], [1, 3, 4], [1, 3, 4, 5]]
config_ = config(trainbase, 'eng', testbase, validbase, lex, phr, cdep, logprefix)
word2vec.rundown_config(config_)
process_qc.datasetConstructRundown_config(config_, dep)
cnn_model.structure_script(config_, filter_hss, hasmlphidden)
cdep = False
dep = True
phr = False
lex = False
hasmlphidden = False
logprefix = '../exp/qc_f_trans_dcnn'
filter_hss = [[3], [3, 4], [3, 4, 5]]
config_ = config(translate_base, 'ch', testbase, validbase, lex, phr, cdep, logprefix)
word2vec.rundown_config(config_)
process_qc.datasetConstructRundown_config(config_, dep)
cnn_model.structure_script(config_, filter_hss, hasmlphidden)
cdep = False
dep = False
phr = False
lex = False
hasmlphidden = False
logprefix = '../exp/qc_f_trans_cnn'
filter_hss = [[3], [3, 4], [3, 4, 5]]
config_ = config(translate_base, 'ch', testbase, validbase, lex, phr, cdep, logprefix)
word2vec.rundown_config(config_)
process_qc.datasetConstructRundown_config(config_, dep)
cnn_model.structure_script(config_, filter_hss, hasmlphidden)
"""