Navigating Sarcasm in Multilingual Text: An In-Depth Exploration and Evaluation

Abstract

Sarcasm means using words when you say something opposite from what you want to say, either to irritate someone, offend them, or just for fun. Detecting sarcasm in multiple languages is yet a challenging area of research. Identifying and understanding sarcasm in social media indicates people's thoughts about specific topics, news, and products. Many articles have been published on sarcasm detection using deep learning and machine learning methods. Moreover, very few systematic reviews have been conducted in this research area. This paper systematically reviews existing Artificial intelligence (A.I.) techniques in sarcastic text detection in different languages. The studies emphasized that in recent literature, machine learning and deep learning, especially recurrent neural networks (RNN), are the most commonly used techniques for sarcasm detection. Twitter is the most frequently used source and accuracy for performance measures. The articles covered in the literature survey also include sarcasm detection from social text, books, and code-mixed text, among other datasets. Finally, this paper briefly discusses the challenges in sarcasm detection and future research in this area.