Intelligent Data Analysis

14Sarcasm Detection Algorithms Based on Sentiment Strength

Pragya Katyayan and Nisheeth Joshi

Department of Computer Science, Banasthali Vidyapith, Rajasthan, India

14.1 Introduction

Sentiments have become a puzzle to be solved these days. Verbal or written expressions of sentiments are tough to comprehend because of the innovative ways people have been adapting in order to express them. Where sentiments used to be a binary value earlier with just positive and negative values to look for, the advent of sarcasm has made the idea a little more explicit. Sarcasm is when someone decides to use words of opposite meaning to what he/she is actually feeling. Sarcasm is the new trendsetter and is so widely used and appreciated that those who do not know it have started to learn it. So, the text we come across in our day-to-day lives, be it on Amazon reviews or Twitter feeds or maybe the daily news headlines are a carrier of sarcasm, in some way or the other. If we wish to detect the sentiment values accurately, we need an algorithm that detects the types of sarcastic expressions along with the positive and negative emotions. According to linguistic Camp E. [1], broadly, sarcasm is of four types: propositional, embedded, “like”-prefixed, and illocutionary. Hyperbole, a type of embedded sarcasm is considered a strong marker of sarcasm. It is recognized when a sentence has both positive and negative polarity. The presence of these contradicting sentiment values is a pointer of hyperbolic ...