class Preprocessor(object):
def __init__(self, configs):
self.ltp_util = LtpUtil(configs)
def process(self, stmt):
"""
对句子进行初始化处理，包括分词、依存句法分析、情感分析等等。
"""
# 分词和词性标注
seg, pos = self.cut(stmt.text)
arcs = self.dependency_parse(seg, pos)
stmt.set_words(list(seg))
stmt.set_pos(list(pos))
stmt.set_arcs(arcs)
stmt.set_emotion(self.emotion_analysis(stmt.text))
return stmt
def cut(self, text, HMM=True):
seg = self.ltp_util.Segmentor(text)
pos = self.ltp_util.Postagger(seg)
return seg, pos
def dependency_parse(self, seg, pos):
# 调用pyltp进行依存句法分析
arcs = self.ltp_util.Parser(seg, pos)
parse_result = []
for item in list(arcs):
parse_result.append((item.head, item.relation))
return parse_result
def emotion_analysis(self, text):
# 返回情感分析结果
return None
class LtpTreeBuilder(object):
def __init__(self, configs):
self.ltp_util = LtpUtil(configs)
def build(self, sentence):
words = self.ltp_util.Segmentor(sentence)
postags = self.ltp_util.Postagger(words)
arcs = self.ltp_util.Parser(words, postags)
head_index = -1
for i in range(len(arcs)):
if arcs[i].head == 0: head_index = i + 1
tree = LtpTree(head_index, 'HED', 0, postags[head_index - 1],
words[head_index - 1])
for i in range(len(arcs)):
if i + 1 != head_index:
p_tree = LtpTree(i + 1, arcs[i].relation, arcs[i].head,
postags[i], words[i])
tree.addChild(p_tree)
return tree
def __init__(self, configs):
self.ltp_util = LtpUtil(configs)