I have a tfidfvectorizer which was fitted on english text data to predict sentiment of english calls. The task is to convert this in spanish languages. I want to use the weights of this tfidfvectorizers and wants to convert features from english to Spanish for e.g. "thank you" becomes "gracias" and use the old weights. So essentially I want to use the same tfidf vectorizer but with modified feature names. Can someone suggest some way to do that in Python.
Edit: I have already converted features from English to Spanish and have trained tfidf on english text. I need is a way to build tfidf using old weights and new features without using fit function or converting all text to spanish.