我可以在 python 中使用 NLTK 从 Spacy 依赖树中找到主题吗？

Question

我想从使用Spacy 的句子中找到主题。下面的代码运行良好并提供了一个依赖关系树。

import spacy
from nltk import Tree

en_nlp = spacy.load('en')

doc = en_nlp("The quick brown fox jumps over the lazy dog.")

def to_nltk_tree(node):
    if node.n_lefts + node.n_rights > 0:
        return Tree(node.orth_, [to_nltk_tree(child) for child in node.children])
    else:
        return node.orth_


[to_nltk_tree(sent.root).pretty_print() for sent in doc.sents]

从这个依赖树代码中，我能找到这句话的主语吗？

Answer 1

我不确定您是否想使用 nltk 解析树编写代码（请参阅 How to identify the subject of a sentence? ）。但是，spacy 也使用 word.dep_ 属性的“nsubj”标签生成它。

import spacy
from nltk import Tree

en_nlp = spacy.load('en')

doc = en_nlp("The quick brown fox jumps over the lazy dog.")

sentence = next(doc.sents) 
for word in sentence:
...     print "%s:%s" % (word,word.dep_)
... 
The:det
quick:amod
brown:amod
fox:nsubj
jumps:ROOT
over:prep
the:det
lazy:amod
dog:pobj

提醒，可能存在多个更复杂的情况。

>>> doc2 = en_nlp(u'When we study hard, we usually do well.')
>>> sentence2 = next(doc2.sents)
>>> for word in sentence2:
...     print "%s:%s" %(word,word.dep_)
... 
When:advmod
we:nsubj
study:advcl
hard:advmod
,:punct
we:nsubj
usually:advmod
do:ROOT
well:advmod
.:punct

Answer 2

与 Leavesof3 一样，我更喜欢将 spaCy 用于这种目的。它具有更好的可视化，即

主题将是具有依赖属性“nsubj”或“普通主题”的单词或短语（如果您使用名词分块）

You can access displaCy (spaCy visualization) demo here

Answer 3

试试这个：

import spacy
import en_core_web_sm
nlp = spacy.load('en_core_web_sm')
sent = "I need to be able to log into the Equitable siteI tried my username and password from the AXA Equitable site which worked fine yesterday but it won't allow me to log in and when I try to change my password it says my answer is incorrect for the secret question I just need to be able to log into the Equitable site"
nlp_doc=nlp(sent)
subject = [tok for tok in nlp_doc if (tok.dep_ == "nsubj") ]
print(subject)