SARCASM DETECTION
Neha Verma1, Ayushi Gupta2, Ms. Renuka Sharma(Assistant Professor ,CSE)
Sahil Bhat3, Madhur Jain4 NIET , Greater Noida U.P
(Computer Science and Engineering, NIET Greater Noida, U.P)
ABSTRACT - The methods and process of detecting sarcasm, as well as the comparison of results from various models and datasets, are discussed in this paper. The term "sarcasm" refers to phrases that convey a meaning that is in opposition to the intended meaning. Recently, NLP has become a subject of great interest for researchers due to its captivating nature. One aspect of NLP that has garnered particular attention is sarcasm detection. This process, similar to sentiment analysis, uses mathematical methods to classify the tone of a text or phrase, determining whether it contains sarcastic elements or not.
On social media, people use sarcasm to covertly communicate their opinions and more intense feelings. Since it necessitates a significant amount of background knowledge, the rhetoric of irony is a subfield of sentiment analysis that is indistinguishable using conventional sentiment analysis methods. The primary aim of present-day sarcasm detection methods is to scrutinize the textual material of sarcasm through a range of natural language processing approaches. Sarcasm involves using language in a derisive or satirical way to ridicule a person or thing. In sarcasm, ridicule or satire is employed harshly, often coarsely, and with disdain for unfavorable consequences. It can be challenging to decipher the true meaning of a sentence in code-mixed language because there aren't enough sarcastic cues. Therefore, the sarcasm detection research that we suggest in this paper uses a variety of techniques and their improvements. This section has examined the Glove Vector[1], Fast Text, Bag-of-Words, Term Frequency-Inverse Document Frequency, (Continuous Bag of Words/Skip Gram), and BERT feature extraction techniques. We use mockery that is past identification to illustrate significant datasets, tactics, patterns, problems, and tasks. To aid researchers in related fields to grasp the cutting-edge practices in sarcasm detection, our research presents succinct tables of sarcasm datasets, features utilized for sarcasm detection, and techniques for their extraction.We performed experiments on various datasets that are accessible to the public, and the outcomes indicate that our suggested approach can greatly improve the precision of identifying sarcasm.
Keywords - sarcasm detection, sentimental analysis, deep learning, support vector machine
