def build_E(mode):
key_subtask = 'E'
rows = data_manager.read_data(key_subtask, mode)
ofname = data_manager.fname_gold(key_subtask, mode)
indexer = Indexer(['-2', '-1', '0', '1', '2'])
topic_label = {}
for row in rows:
topic = row[1]
label = row[2]
if topic not in topic_label:
topic_label[topic] = np.zeros(indexer.size())
topic_label[topic][indexer.idx(label)] += 1
fobj = open(ofname, 'w')
for topic, labels in topic_label.items():
n_sample = np.sum(labels)
dist = labels / n_sample
fobj.write('%s\t%s\n' %
(topic, '\t'.join(map(lambda k: '%.12f' % k, dist))))
fobj.close()
def build_C(mode):
key_subtask = 'C'
rows = data_manager.read_data(key_subtask, mode)
ofname = data_manager.fname_gold(key_subtask, mode)
fobj = open(ofname, 'w')
for row in rows:
fobj.write('%s\t%s\t%s\n' % (row[0], row[1], row[2]))
fobj.close()