I'm trying to write a script in Python using nltk which changes a sentence from second person to first person. Example: the sentence
I went to see Avatar and you came with me
should become
You went to see Avatar and I came with you
Is there a built-in function in nltk that does this?
There shouldn't be too many forms of personal and possessive pronouns in English. If you create a dictionary of correspondence between 1st and 2nd person forms, you can then tokenize the original sentence and replace the words that are in the dictionary:
forms = {"am" : "are", "are" : "am", 'i' : 'you', 'my' : 'yours', 'me' : 'you', 'mine' : 'yours', 'you' : 'I', 'your' : 'my', 'yours' : 'mine'} # More?
def translate(word):
if word.lower() in forms: return forms[word.lower()]
return word
sent = 'You went to see Avatar, and I came with you.'
result = ' '.join([translate(word) for word in nltk.wordpunct_tokenize(sent)])
print(result.caputalize())
# I went to see avatar , and you came with i .
Because of the ambiguity of you you probably cannot get any better results.