I have a dataset containing raw facebook posts and comments. What I would like to do is to perform sentiment analysis with Python 3 (NTLK ?) in order to label each post and each comment against some categories (a sort of clustering in unsupervised mode). The problem is that I have no idea how to do something similar.
I accept suggestions Thank you
You can use vaderSentiment which is a python package to perform unsupervised English Sentiment Analysis using dictionary and rules. There is some example on their github. This option might be more effective than an unsupervised clustering.