def mentionEntityVocabularyForTest():
train_entity_friendlyname = files_handle.read_set(
"..\\data\\entity\\graphquestions.training.entityFriendlyname")
#test_questions_entity_friendlyname()
#mentionEntityVocabularyForTest()
def test_questions_entity_friendlyname():
test_entity_friendlyname = files_handle.read_set(
"..\\data\\entity\\graphquestions.testing.entityFriendlyname")
train_entity_friendlyname = files_handle.read_set(
"..\\data\\entity\\graphquestions.training.entityFriendlyname")
test_entity_friendlyname_remain = test_entity_friendlyname - train_entity_friendlyname
print(len(test_entity_friendlyname_remain))
# print(test_entity_friendlyname_remain)
test_entity_friname = entity_friendlyname(test_entity_friendlyname_remain)
miss_after_alias = graphentityalias_miss(test_entity_friname)
test_entity_friname_miss_after_alias = entity_friendlyname(
miss_after_alias)
miss_after_alias_name = graphentitynamemiss(
test_entity_friname_miss_after_alias)
test_entity_friname_miss_after_alias_name = entity_friendlyname(
miss_after_alias_name)
miss_after_alias_name_clueweb = cluewebmiss(
test_entity_friname_miss_after_alias_name)
files_handle.write_set(
miss_after_alias_name_clueweb,
"..\\data\\entity\\graphquestions.testing.entityFriendlynamemiss")
def compare_NP_mention():
position_question_posword = test_easy_mention_position()
np_mention_indexrange = read_dict_mention_indexrange(
"..\\data\\test\\test.easy.ques.np.index.range")
not_in_np_ques_mention_pos_indexrange = dict()
not_in_np_ques_mention_pos = dict()
for position in position_question_posword:
for question_posword in position_question_posword[position]:
mention_indexranges = np_mention_indexrange[question_posword.split(
"###")[0]]
if position not in mention_indexranges:
not_in_np_ques_mention_pos_indexrange[
question_posword.split("###")
[0]] = position + "###" + "###".join(mention_indexranges)
not_in_np_ques_mention_pos[question_posword.split("###")
[0]] = position
# print("not in\t"+question_posword.split("###")[0])
ques_position_pos_equal = train_mention_pos_equal_position()
questions_unhittes = read_set("../data/test/test.easy.ques.np_not_hit")
for ques in questions_unhittes:
position_right = not_in_np_ques_mention_pos[ques]
if ques in ques_position_pos_equal:
if position_right in ques_position_pos_equal[ques]:
print("in:\t" + ques)
ques_frname_in_position = train_frname_in_test_position()
for ques in questions_unhittes:
position_right = not_in_np_ques_mention_pos[ques]
if ques in ques_frname_in_position:
if position_right in ques_frname_in_position[ques]:
print("in:\t" + ques)
# write_dict_str(not_in_np_ques_mention_pos_indexrange,"..\\data\\test\\test.easy.ques.not_in_np_ques_mention_pos_indexrange")
range_num = dict()
# for question in not_in_np_ques_mention_pos_indexrange:
# position=not_in_np_ques_mention_pos_indexrange[question].split("###")[0]
# # print(position)
# range_position=0
# if "\t" in position:
# position_start=position.split("\t")[0]
# position_end=position.split("\t")[1]
# range_position=int(position_end)-int(position_start)+1
# else:
# range_position=1
# if range_position in range_num:
# range_num[range_position]=range_num[range_position]+1
# else:
# range_num[range_position] = 1
# print(range_num)
return not_in_np_ques_mention_pos_indexrange
def friendlyname_entity():
train_entity_friendlyname = files_handle.read_set(
"..\\data\\entity\\graphquestions.training.entityFriendlyname")
train_friendlyname_entity = friendlyname_entity_exchange(
train_entity_friendlyname)
return train_friendlyname_entity
import time
from entity_linker.files_handle import read_posques_posword, write_dict_dict_dict, read_set
from entity_linker.name_entity_files_handle import friendlyname_entity, alias_entity, name_entity, clueweb_name_entity
print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())))
train_friendlyname_entity = friendlyname_entity()
print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())))
graphq_alias_entity = alias_entity()
print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())))
graph_nameentity = name_entity()
print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())))
name_entitygraphq_pro_clueweb = clueweb_name_entity()
print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())))
train_pos_combine = read_set(
"../data/test/intersect.train.test.easy.mention.pos.composition")
print(train_pos_combine)
def posword_wordlist(posword):
word_list = list()
for pos_word in posword:
word = pos_word.split("\t")[1].lower()
word_list.append(word)
return word_list
def posword_poslist(posword):
pos_list = list()
for pos_word in posword:
pos = pos_word.split("\t")[0].lower()
pos_list.append(pos)