Enhanced lexicon E-SLIDE framework for efficient sentiment analysis

Idioms are multi-word non-compositional expressions whose meaning is different from the underlying meaning and thereby posing a significant challenge in the interpretation. Found in all the languages, idioms beautify a language but complicate the Sentiment Analysis task. The Extended Sentiment Lexicon of IDiomatic Expressions (ESLIDE)1, created for this work, is an extension of the state-of-the-art lexicon of idioms-SLIDE which comprised of five thousand frequently occurring idioms estimated from a large corpus of English language. The contribution of idioms as features in sentiment analysis task is examined in this paper. The classifiers—Naive Bayes, Multinomial Naive Bayes and Decision Trees used in this work are evaluated using accuracy, precision and recall, noticing a slight increase in the performance over all three sentiment classes viz. positive, negative and neutral improving the baseline results by six percent points. © 2021, Bharati Vidyapeeth's Institute of Computer Applications and Management.