SARCASM DETECTION USING DEEP LEARNING

Abstract

Even for humans, it can be difficult to recognize sarcasm, which is an important part of communication. To attract readers’ attention, sarcasm is frequently used in newspaper headlines. Although headlines usually include irony, readers frequently miss it, misinter preting the news and spreading misinformation to friends, coworkers, and other people. As a result, it is more important than ever to have a system that can automatically and reliably recognize sarcasm. In order to build sarcasm detectors, we employ neural net works, and we investigate how a computer may learn sarcastic patterns. The sequences that are fed into our project might be ironic or not. From collections of news headlines, these sequences were created. Our classifiers are evaluated for accuracy. Our approach is effective at identifying the difference between remarks that are sarcastic and those that are not. Sarcasm identification is a method for spotting expressions that mean the exact oppo site of what they mean to say. Systems for sentiment analysis that rely on emotion recog nition face a substantial barrier due to the metaphorical character of sarcasm. Natural language processing (NLP) is a highly specialised field that focuses on sarcasm identifi cation rather than sentiment analysis across many domains. Sarcasm detection in online forums is one such application. Though they differ slightly, sarcasm detection and sentiment analysis are closely linked. Sarcasm recognition is a specialised study topic in the science of NLP that aims to as certain if a particular text or phrase is sardonic. In a recent study, we investigated and contrasted several methods for sarcasm detection. This study aims to tackle the problem of analysing big datasets while preserving the effectiveness of Deep Learning models. We seek elegant answers that provide efficient computing without compromising effective analysis.