A DEEP LEARNING APPROACH TO SARCASM DETECTION FROM COMPOSITE TEXTUAL DATA

• Autor(a) principal: Khan, Amit
• Data de Publicação: 2022
• Outros Autores: DIPANKAR MAJUMDAR, BIKROMADITTYA MONDAL, SOUMEN MUKHERJEE
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: INFOCOMP: Jornal de Ciência da Computação
• Texto Completo: https://infocomp.dcc.ufla.br/index.php/infocomp/article/view/2266

Sarcasm is a means of conveying a bad attitude through social media platforms by utilizing positive or exaggerated positive terms. The last decade witnessed sarcasm detection to become a highly phenomenal topic of research; however the task of automated detection of sarcastic comments in a text remains an elusive problem. Sarcasm detection has eventually become a considerably significant task in the domain of sentiment classification. Without properly detecting the sarcasm from textual comments, sentiment classification remains incomplete and may lead to wrongful conclusion and decision. In this paper, we present a recurrent neural network (RNN)-based bidirectional long-short term memory (Bi-LSTM) network for sarcasm detection.The proposed technique has been applied to a combined dataset which is produced form news headline sarcasm dataset and news headline sarcasm version 2. Results of our technique renders enhanced performance over the existing technique found in literature.