A Hybrid Feature Selection Method to Detect Sarcasm in Telugu Text Leveraging AdaBoost Classifier.

Abstract

Automatic sarcasm detection refers to the process of predicting sarcasm inside text. The ubiquity and complexities of sarcasm in sentiment-laden text represent a crucial stage in numerous sentiment analysis endeavours. Every day, individuals worldwide utilize social media platforms to disseminate their opinions, experiences, and suggestions. Sarcasm is often employed in newspaper headlines to capture readers' attention. Consequently, the necessity for a system capable of automatically and consistently detecting sarcasm has become paramount. Low-resourced and morphologically rich languages, such as Telugu, have garnered considerable attention from scholars. In Telugu sarcasm can be represented using proverbs, repeating the same word multiple times and also mentioning ellipses. In Telugu, sarcasm can be conveyed through proverbs, the repetition of words, and the use of ellipses. Leverage neural networks to create sarcasm detectors and investigate the methods by which a machine might learn sarcastic patterns. This research analysed hybrid feature selection utilizing the Gini index and entropy in conjunction with the AdaBoost machine learning classifier to enhance its performance.