def preprocess_data_to_merge(input_standoff_folder_gold,
output_conll_folder_gold, output_conll_file_gold,
input_standoff_folder_pred,
output_conll_folder_pred, output_conll_file_pred):
# if not os.path.exists(output_conll_folder_gold):
# os.makedirs(output_conll_folder_gold)
# else:
# shutil.rmtree(output_conll_folder_gold)
# anntoconll_wlp.covert_standoff_to_conll(input_folder_main= input_standoff_folder_gold, output_folder = output_conll_folder_gold)
anntoconll_wlp.convert_standoff_conll_single_file(
input_standoff_folder_gold, output_conll_folder_gold,
output_conll_file_gold)
list_of_test_files_stand_off = Read_Files_in_Input_Folder(
output_conll_folder_gold)
for file_name in list_of_test_files_stand_off:
file_values = file_name.split("/")
protocol_name = file_values[-1]
conll2standoff.process(file_name, input_standoff_folder_gold)
copy_text_files(input_standoff_folder_gold, input_standoff_folder_pred)
anntoconll_wlp.convert_standoff_conll_single_file(
input_standoff_folder_pred, output_conll_folder_pred,
output_conll_file_pred)
def preprocess_data(input_standoff_folder_train, output_conll_folder_train,
output_conll_file_train, input_standoff_folder_test,
output_conll_folder_test, output_conll_file_test):
anntoconll_wlp.convert_standoff_conll_single_file(
input_standoff_folder_train, output_conll_folder_train,
output_conll_file_train)
anntoconll_wlp.convert_standoff_conll_single_file(
input_standoff_folder_test, output_conll_folder_test,
output_conll_file_test)