Sarcasm Detection on Twitter using Machine and Pragmatic classifier

Authors

  • Deepti Jeetu Janjani, Dr. Deepali Vora, Dr. Abhishek Anurag

Abstract

Sarcasm is major entity of social media .Sarcasm is a form of language, where person
mean the opposite of what is being implied. Sarcasm is a pervasive linguistic phenomenon at social
sites that express subjective and deeply- felt opinions. Twitter has provided a platform where in all
peoples are free to express their views regarding subject or a person. Twitter itself has millions of
active users, with increase in number of users, data at twitter also increases which is Big Data. With
growing era of internet and India is blessed with many languages, detection of slang words is itself a
challenge, hence there is no particular way of defining sarcasm. The already available corpus of
positive and negative sentiments may not prove to be that accurate in detecting sarcasm as there is
rapid growth in day to day use of emoticons, it may change the polarity of statement. It is quite a
challenge to detect sarcastic content. There is been a necessity to work for emoticons based sarcasm
detection along with text rather than only based on sentence approach of detecting sarcasm. Our
work ensembles the use of two pedagogy –voted ensemble classifier and random forest classifier.
The one that are already existing approaches which uses predefined corpus of positive and negative
sentiments to train the classifiers, unlike that here it is used seeding approach to train the corpus,
whereas to detect emoticons we use pragmatic classifiers.

Published

2020-10-16

Issue

Section

Articles