def test_parser(config, best_model, data_types):
base_dir = config['data']['base_dir']
features = ['sents', 'predicted_pos', 'predicted_stag', 'arcs', 'rels']
base_command = 'python graph_parser_main.py test --get_accuracy'
model_info = ' --model {}'.format(best_model)
model_info += ' --metrics {}'.format(config['parser']['scores']['metrics'])
for data_type in data_types:
inputs = {}
output_file = os.path.join(base_dir, 'predicted_arcs', '{}.txt'.format(data_type))
inputs[6] = output_file
if not os.path.isdir(os.path.dirname(output_file)):
os.makedirs(os.path.dirname(output_file))
output_info = ' --predicted_arcs_file {}'.format(output_file)
output_file = os.path.join(base_dir, 'predicted_rels', '{}.txt'.format(data_type))
inputs[7] = output_file
if not os.path.isdir(os.path.dirname(output_file)):
os.makedirs(os.path.dirname(output_file))
output_info += ' --predicted_rels_file {}'.format(output_file)
inputs_greedy = {}
output_file = os.path.join(base_dir, 'predicted_arcs_greedy', '{}.txt'.format(data_type))
inputs_greedy[6] = output_file
if not os.path.isdir(os.path.dirname(output_file)):
os.makedirs(os.path.dirname(output_file))
output_info += ' --predicted_arcs_file_greedy {}'.format(output_file)
output_file = os.path.join(base_dir, 'predicted_rels_greedy', '{}.txt'.format(data_type))
inputs_greedy[7] = output_file
if not os.path.isdir(os.path.dirname(output_file)):
os.makedirs(os.path.dirname(output_file))
output_info += ' --predicted_rels_file_greedy {}'.format(output_file)
test_data_dirs = map(lambda x: os.path.join(base_dir, x, '{}.txt'.format(data_type)), features)
test_data_info = ' --text_test {} --jk_test {} --tag_test {} --arc_test {} --rel_test {}'.format(*test_data_dirs)
complete_command = base_command + model_info + output_info + test_data_info
subprocess.check_call(complete_command, shell=True)
#output_conllu(os.path.join(base_dir, config['data']['split'][data_type]), os.path.join(base_dir, config['data']['split'][data_type]+'_arc_rel'), inputs)
output_conllu(os.path.join(base_dir, config['data']['split'][data_type]), os.path.join(base_dir, config['data']['split'][data_type]+'_arc_rel_greedy'), inputs_greedy)
def test_stagger(config, best_model, data_types):
base_dir = config['data']['base_dir']
base_command = 'python bilstm_stagger_main.py test'
model_info = ' --model {}'.format(best_model)
for data_type in data_types:
output_file = os.path.join(base_dir, 'predicted_stag',
'{}.txt'.format(data_type))
if not os.path.isdir(os.path.dirname(output_file)):
os.makedirs(os.path.dirname(output_file))
output_info = ' --save_tags {} --get_accuracy'.format(output_file)
test_data_info = ' --text_test {} --jk_test {} --tag_test {}'.format(
os.path.join(base_dir, 'sents', '{}.txt'.format(data_type)),
os.path.join(base_dir, 'gold_stag', '{}.txt'.format(data_type)),
os.path.join(base_dir, 'gold_stag', '{}.txt'.format(data_type)))
complete_command = base_command + model_info + output_info + test_data_info
subprocess.check_call(complete_command, shell=True)
output_conllu(
output_file,
os.path.join(base_dir, config['data']['split'][data_type]),
os.path.join(base_dir,
config['data']['split'][data_type] + '_stag'))