Sentiment Analysis on Telugu–English Code-Mixed Data

Abstract

AbstractWith the increase of communication over social media, in a multilingual country like India, people tend to use more than one language in day-to-day communication and also on social media platforms in order to showcase their linguistic proficiency. For instance, combining Telugu and English or Tamil and English in the same sentence is commonly observed. This is called code-mixing. Code-mixed text demands a different level of processing, and in this paper, we have attempted to extract sentiment from the Telugu–English code-mixed sentences. We classify the polarity of the code-mixed sentences into positive and negative sentiments classes using lexicon-based approach and also using machine learning approaches, namely naïve Bayes and support vector machine classifiers. We achieved an accuracy of 82% and 85%, respectively.KeywordsNatural language processingSentiment extractionCode-mixed dataMachine learningLexiconsNaïve BayesSVM