如何使用 spaCy 从句子中删除实体？答案

【问题标题】：How to remove an entity from a sentence with spaCy?如何使用 spaCy 从句子中删除实体？
【发布时间】：2021-02-23 00:29:44
【问题描述】：

如何使用 spaCy 从句子中删除实体？我想随机删除 ORP、GPE、Money、Ordinal 或 Percent 实体。例如，

唐纳德·约翰·特朗普[人]（生于 1946 年 6 月 14 日）[日期] 是美国第 45 任[序数] 和现任总统 [GPE]。在进入政界之前，他是一名商人和电视名人。

现在如何从这句话中删除某个实体？在此示例中，该函数选择删除第 45 个序数实体。

>>> sentence = 'Donald John Trump (born June 14, 1946) is the 45th and current president of the United States. Before entering politics, he was a businessman and television personality.'
>>> remove(sentence)
45th

【问题讨论】：

标签： python nlp nltk spacy

【解决方案1】：

请尝试SpacyNER 和np.random.choice：

import spacy
nlp = spacy.load("en_core_web_md")

sentence = 'Donald John Trump (born June 14, 1946) is the 45th and current president of the United States. Before entering politics, he was a businessman and television personality.'
doc = nlp(sentence)

ents = [e.text for e in doc.ents if e.label_ in ("NORP", "GPE", "MONEY", "ORDINAL","PERCENT")]
remove = lambda x: str(np.random.choice(x))
# expected output
remove(ents)
'45th'

您是否希望从句子文本中删除随机实体：

def remove_from_sentence(sentence):
    doc = nlp(sentence)
    with doc.retokenize() as retokenizer:
        for e in doc.ents:
            retokenizer.merge(doc[e.start:e.end])
    tok_pairs = [(tok.text, tok.whitespace_) for tok in doc]
    ents = [e.text for e in doc.ents if e.label_ in ("NORP", "GPE", "MONEY", "ORDINAL","PERCENT")]
    ent_to_remove = remove(ents)
    print(ent_to_remove)
    tok_pairs_out = [pair for pair in tok_pairs if pair[0] != ent_to_remove]
    return "".join(np.array(tok_pairs_out).ravel())

remove_from_sentence(sentence)

the United States
'Donald John Trump (born June 14, 1946) is the 45th and current president of . Before entering politics, he was a businessman and television personality.'

如果有不清楚的地方请追问。

【讨论】：