Sarcasm Detection on Twitter using Deep Handcrafted Features

Authors

  • Hasnat Saleem Department of Computer Science, NFC-IET, Multan, 60000, Pakistan.
  • Ahmad Naeem Department of Computer Science, NFC-IET, Multan, 60000, Pakistan.
  • Kamran Abid Department of Computer Science, NFC-IET, Multan, 60000, Pakistan.
  • Naeem Aslam Department of Computer Science, NFC-IET, Multan, 60000, Pakistan.

Keywords:

Deep learning, Sarcasm detection, Social media, Feature engineering , Twitter

Abstract

The recent advancement of social media has greatly impacted people's daily lives. People nowadays express their emotions through social media. Twitter is the most utilized social media platform for people to share information. As social media popularity is increasing a lot of information available on the internet becomes dubious and misleading. The sarcastic text is a true lie spread through social media and it is a statement that is different from the actual message. It is very challenging to recognize sarcasm from social media manually. Therefore, the detection of sarcasm is essential from social media using an advanced automated system based on deep learning methods. In this study, we have proposed a novel method for the detection of sarcasm. In this study, BoW, TF-IDF, and word embeddings are used to detect the prominent features from the text and a long-short memory (LSTM) network for identifying sarcastic remarks in a given corpus. The publicly available Twitter dataset is used in this study which is based on sarcasm, irony, and regular tweets. To evaluate the methods, we have used recall, precision, F1, and accuracy score as the evaluation parameters. The proposed model achieved a 99.01% accuracy for the detection of sarcasm on social media.

Downloads

Published

2023-03-29

How to Cite

Hasnat Saleem, Ahmad Naeem, Kamran Abid, & Naeem Aslam. (2023). Sarcasm Detection on Twitter using Deep Handcrafted Features. Journal of Computing & Biomedical Informatics, 4(02), 117–127. Retrieved from https://jcbi.org/index.php/Main/article/view/128