class EventVector:
def __init__(self):
self.mongo_operation = MongoOperation()
self.ltp_parser = LtpParser()
self.words_embedding_file = "../Data/Tencent_AILab_ChineseEmbedding/Tencent_AILab_ChineseEmbedding.txt"
self.wv_from_text = KeyedVectors.load_word2vec_format(
self.words_embedding_file, binary=False, limit=100000)
self.wv_from_text.init_sims(replace=True)
# 从数据库中得到事件三组元按“,”划分的事件，并不重复
def get_event_from_triple(self):
event_triple_sets = self.mongo_operation.event_db_get()
event_sets = list()
for event_triple in event_triple_sets:
event = event_triple.split(',')[0]
if event not in event_sets:
event_sets.append(event)
event = event_triple.split(',')[2]
if event not in event_sets:
event_sets.append(event)
return event_sets
# 得到事件的向量（词向量平均得到）
def get_event_vectors(self, event_sets):
events_list = list()
for event in event_sets:
event_dict = dict()
words_list = self.ltp_parser.get_words_by_pyltp(event)
# print(words_list)
# 建立长度向量长度的0向量
vector_sum = [
0 for index in range(self.wv_from_text.wv.syn0[0].shape[0])
]
for word in words_list:
vector = self.wv_from_text[word]
# print(vector)
vector_sum = list(
numpy.array(vector_sum) + numpy.array(vector))
# print(vector_sum)
event_vector = [i / len(words_list) for i in vector_sum]
event_dict["event"] = event
event_dict["vector"] = event_vector
events_list.append(event_dict)
return events_list
def __init__(self):
# fd = open(self.EVENT_RELATIONS_LIST_FILE_NAME, 'r')
# self.event_triple_sets = fd.readlines()
db = MongoOperation()
self.event_triple_sets = db.event_db_get()
print('\n'.join(self.event_triple_sets))