def __init__(self, bigrams=False, encoding_type='bmeso'):
super().__init__()
self.bigrams = bigrams
if encoding_type=='bmeso':
self.encoding_func = lambda x:x
elif encoding_type=='bio':
self.encoding_func = bmeso2bio
elif encoding_type == 'bioes':
self.encoding_func = lambda words: iob2bioes(iob2(words))
else:
raise RuntimeError("Only support bio, bmeso, bioes")
def __init__(self,
encoding_type: str = 'bio',
lower: bool = False,
word_shape: bool = False):
if encoding_type == 'bio':
self.convert_tag = iob2
elif encoding_type == 'bioes':
self.convert_tag = lambda words: iob2bioes(iob2(words))
else:
raise ValueError("encoding_type only supports `bio` and `bioes`.")
self.lower = lower
self.word_shape = word_shape
def __init__(self, encoding_type: str = 'bio', lower: bool = False, word_shape: bool=False):
"""
:param: str encoding_type: target列使用什么类型的encoding方式，支持bioes, bio两种。
:param bool lower: 是否将words小写化后再建立词表，绝大多数情况都不需要设置为True。
:param boll word_shape: 是否新增一列word shape，5维
"""
if encoding_type == 'bio':
self.convert_tag = iob2
elif encoding_type == 'bioes':
self.convert_tag = lambda words: iob2bioes(iob2(words))
else:
raise ValueError("encoding_type only supports `bio` and `bioes`.")
self.lower = lower
self.word_shape = word_shape