用斯佩西那识别人，让人一个字？

Question

我想用Spacy NER来识别那个人，然后把它写成一个词。

我的数据集如下所示：

text     
use your superpowers
vote for Barack Obama
vote for Marine Le Pen
play with Michael Jordan
support the supporters

我希望我的最后输出如下所示：

text     
use your superpowers
vote for Barack_Obama
vote for Marine_Le_Pen
play with Michael_Jordan
support the supporters

这是我到目前为止掌握的代码：

 def get_ner (string):
     nlp = spacy.load("en_core_web_trf")
     doc = nlp(string)
     for token.text in doc:
         if token.ents=="Person":
         s= ent['start']
         e= ent['end']
         txt = txt[:s] + txt[s:e+1].replace(' ', '_') + txt[e:]
     return txt

 df['text']= df.text.apply(get_ner)

当我使用上面的代码时，我会收到一条错误消息。

AttributeError: name 'token' is not defined

Answer 1

如果使用Spacy，您的代码应该是：

nlp = spacy.load('en_core_web_trf')

def get_ner(txt):
    doc = nlp(txt)
    for ent in doc.ents:
        if ent.label_ == 'PERSON':
            s = ent.start_char
            e = ent.end_char
            txt = txt[:s] + txt[s:e+1].replace(' ', '_') + txt[e:]
    return txt

df['text'] = df['text'].apply(get_ner)

输出：

>>> df
                       text
0      use your superpowers
1     vote for Barack_Obama
2    vote for Marine_Le_Pen
3  play with Michael_Jordan
4    support the supporters