Twitter sentiment analysis on political tweets

Twitter is one of the most important social media platforms that is used to keep oneself updated about any political or any news. Twitter generally has a word limit, which also limits the words to be detected in the tweets. Twitter has developed into one of the primary platforms for political discourse and public policy discussions. This paper proposes a Twitter sentiment analyser for political tweets using Machine Learning (ML) and Natural Language Processing (NLP). This paper envisions a model which performs sentiment analysis on tweets discussing a particular political or policy topic from a particular date and performs sentiment analysis on that data. The system is trained with a dataset of 160000 tweets commenting on Indian politics. The system uses NLP for pre-processing and feature extraction. Classification is performed using 3 classifiers namely Random forest, Support Vector Machine (SVM) and Bernoulli Naive Bayes. The highest accuracy obtained was 88.36% using SVM classifier.