Hello Gobi,

It depends in your approach, in general there are two main approaches: statistical approaches and lexicon based approaches. For the first one, you can use machine learning techniques, so you need an annotated corpus to train you classifier. And for the second one, you need an affective lexicon (a list of word with their affective score).

Hi Gobi,

I have written a pretty elaborative answer to your question on another post:

https://www.researchgate.net/post/How_to_identify_positive_negative_and_neutral_polarities_in_subjective_sentences2

Hope this helps.