def combine_comm(self, words):
"""根据词性标注进行普通实体合并
Args:
words: WordUnit list，进行命名实体合并后的words
Returns:
words_combine: WordUnit list，进行普通实体连接后的words
"""
newword = words[0].lemma # 第一个词，作为新词
words_combine = [] # 存储合并后的结果
n = 1
i = 1 # 当前词ID
while i < len(words):
word = words[i]
# 词合并: (前后词都是实体) and (前后词的词性相同 or 前词 in ["nz", "j"] or 后词 in ["nz", "j"])
if (self.is_entity(word.postag)
and self.is_entity(words[i - 1].postag)
and (word.postag in {'nz', 'j'}
or words[i - 1].postag in {'nz', 'j'})):
newword += word.lemma
else:
words_combine.append(WordUnit(n, newword,
words[i - 1].postag)) # 添加上一个词
n += 1
newword = word.lemma # 当前词作为新词
i += 1
# 添加最后一个词
words_combine.append(WordUnit(n, newword,
words[len(words) - 1].postag))
return words_combine
def combine(self, words, netags):
"""根据命名实体的B-I-E进行词合并
Args:
words: WordUnit list，分词与词性标注后得到的words
netags: list，命名实体识别结果
Returns:
words_combine: WordUnit list，连接后的结果
"""
words_combine = [] # 存储连接后的结果
length = len(netags)
n = 1 # 实体计数，从1开始
i = 0
while i < length:
if 'B-' in netags[i]:
newword = words[i].lemma
j = i + 1
while j < length:
if 'I-' in netags[j]:
newword += words[j].lemma
elif 'E-' in netags[j]:
newword += words[j].lemma
break
elif 'O' == netags[j] or (j + 1) == length:
break
j += 1
words_combine.append(
WordUnit(n, newword, self.judge_postag(netags[j - 1])))
n += 1
i = j
else:
words[i].ID = n
n += 1
words_combine.append(words[i])
i += 1
return self.combine_comm(words_combine)
def postag(self, lemmas):
"""对分词后的结果进行词性标注
Args:
lemmas: list，分词后的结果
entity_dict: set，实体词典，处理具体的一则判决书的结构化文本时产生
Returns:
words: WordUnit list，包含分词与词性标注结果
"""
words = [] # 存储句子处理后的词单元
# 词性标注
postags = self.postagger.postag(lemmas)
for i in range(len(lemmas)):
# 存储分词与词性标记后的词单元WordUnit，编号从1开始
word = WordUnit(i+1, lemmas[i], postags[i])
words.append(word)
# self.postagger.release() # 释放
return words
return words_str.rstrip('\n')
def get_lemmas(self):
"""获得句子的分词结果
Returns:
lemmas: str，该句子的分词结果
"""
lemmas = ''
for word in self.words:
lemmas += word.lemma + '\t'
return lemmas.rstrip('\t')
if __name__ == '__main__':
# 中国首都北京
word3 = WordUnit(3, '北京', 'ns', 0, None, 'HED')
word2 = WordUnit(2, '首都', 'ns', 3, None, 'ATT')
word1 = WordUnit(1, '中国', 'ns', 2, None, 'ATT')
words = [] # 句子的词单元
words.append(word1)
words.append(word2)
words.append(word3)
sentence = SentenceUnit(words)
print(sentence.to_string())
print('句子分词结果: ' + sentence.get_lemmas())
print('"首都"的中心词lemma: ' + sentence.words[1].head_word.lemma)
print('句子的中心词: ' + sentence.get_head_word().to_string())