Accuracy enhancement of textual data during sentiment analysis in twitter using multinomial naive bayes

Abstract

Many messages convey opinions about various things, including people, politics, products, services, and even people's emotions and moods. Sentiment analysis has a wide range of uses, including analysing the outcomes of social network events and examining consumer views of goods and services. The popularity of social media sites like Facebook, Twitter, LinkedIn, and Instagram has made it possible for people to express their thoughts, feelings, and views on a wide range of subjects. A lot of research has been done in this area, but accuracy in analysing sentiments can still be enhanced. For this study, we have considered the Kaggle data set to figure out the sentiments used in racist and non-racist tweets. The techniques that are employed in this study are the Naive Bayes algorithm and NLP. The proposed model achieves 94% accuracy.