I'm trying to understand a Key-Bigram extractor's working and I cannot understand what does the following block of code do.

Here is the source code.

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

string = '1 2 3 4 5 6 7 8 9'


def textProcessing(doc):
    Words = []
    doc = nlp(doc)

    for possible_words in doc:
        Words.append([possible_words , [child for child in possible_words.children]])
    
    print(Words)

textProcessing(string)
       

Everything else is workin fine and I understood well, however I can not understand what child for child in possible_words.children does.

CodePudding user response：

I don't know spacy's API very well, being a user of nltk ;)

But, simply put, as possible_words is a Token, its children property returns all tokens that sit in the syntactic unit of the parent token.

Therefore, [child for child in possible_words.children] turns the iterable returned by this property into a list. I'd have written it list(possible_words.children), though.

CodePudding user response：

token.children uses the dependency parse to get all tokens that directly depend on the token in question. In a visualization (try displacy), this will be all the tokens with arrows pointing away from a token; if the word is a verb this could be the subject and any objects, if the word is a noun it could be any adjectives modifying it, for example.

CodePudding user response：

looks like it's a class in Python and children is an element of the "possible_word" variable stored in "spacy" file