def nmf30_feature(data):
data_block_number = int((data.shape[0] - 1) / block_size) + 1
data = process_data(nmf30,
data,
n=2,
tag='topic model',
data_n=data_block_number)
return data
def local_vec_feature(data):
data_block_number = int((data.shape[0] - 1) / block_size) + 1
print("block num for loc vec", data_block_number, data.shape[0])
data = process_data(local_vec,
data,
n=4,
tag='local word2vec',
data_n=data_block_number)
return data
def basic_feature():
data = pd.read_csv('data/{f}'.format(f=data_file), sep=',')
if n > 0:
data = data[start:n]
if data_file != "test.csv":
data = data.drop(['id', 'qid1', 'qid2'], axis=1)
print("basic data is read")
ret = process_data(base_process_thread, data, n=8, tag="base_feature")
return ret
def pos_vec_feature(data):
# data_block_number = int((data.shape[0]-1) / block_size) + 1
data = process_data(pos_vec, data, n=8,
tag='pos vec') #, data_n=data_block_number)
return data
def common_ratio_feature(ret):
ret = process_data(common_ratio_thread, ret, n=6, tag="common_ratio")
return ret
def dist_features(ret):
ret = process_data(dist_thread, ret, n=3, tag='dist_feature')
return ret
def norm_wmd_feature(ret):
ret = process_data(norm_wmd_tread, ret, n=3, tag="norm_wmd_feature")
return ret
def wmd_feature(ret):
ret = process_data(wmd_thread, ret, n=3, tag="wmd_feature")
return ret
