Context-based Sarcasm Detection in Hindi Tweets

• Published in: 2017 Ninth International Conference on Advances in Pattern Recognition (ICAPR) pp. 1 - 6
• Main Authors: Bharti, Santosh Kumar, Babu, Korra Sathya, Raman, Rahul
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.12.2017
• Subjects: Detectors; Hindi; Online News; Sarcastic; Sentiment; Sentiment analysis; Social Media; Tagging; Testing; Training; Tweets; Twitter
• DOI: 10.1109/ICAPR.2017.8593198

Sentiment analysis is the way of finding ones' opinion towards any specific target. Sarcasm is a special type of sentiment which infers the opposite meaning of what people convey in the text. It is often expressed using positive or intensified positive words. Nowadays, posting sarcastic messages on social media like Twitter, Facebook, WhatsApp, etc., became a new trend to avoid direct negativity. In the presence of sarcasm, sentiment analysis on these social media texts became the most challenging task. Therefore, an automated system is required for sarcasm detector in textual data. Many researchers have proposed several sarcasm detection techniques to identify sarcastic text. These techniques are designed to detect sarcasm on the text scripted in English since it is the most popular language in social networking groups. However, parallel research for sarcasm detection on different Asian languages like Hindi, Telugu, Tamil, Urdu, and Bengali are not yet explored. One of the reasons for the less exploration of these languages for sarcastic sentiment analysis is the lack of annotated corpus even though they are popular in a large networked society. In this article, we proposed a context-based pattern i.e. "sarcasm as a contradiction between a tweet and the context of its related news" for sarcasm detection in Hindi tweets. The proposed approach utilized Hindi news as the context of a tweet with in the same timestamp and attained an accuracy of 87 %.