Deep LSTM-RNN with Word Embedding for Sarcasm Detection on Twitter
Sayed Saniya Salim, Agrawal Nidhi Ghanshyam, Darkunde Mayur Ashok, Dungarpur Burhanuddin Mazahir, Bhushan Thakare
Abstract
In the world full of sarcastic people, it’s becoming challenging for the people of 21 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">st</sup> century to detect sarcasm using sentiment analysis efficiently. Sarcasm detection helps us to understand the bitter truth under the sugar coated sentences. It is widely used in various networking sites for understanding the true reviews and taking appropriate actions on the same if needed. Various methods, techniques and algorithms have been applied, although there’s little or much drawback of using the same. For instance, algorithms like logistic regression has been used to detect sarcasm, which has a drawback, can’t be used for continuous datasets. In our paper, we will be discussing about the approach we found appropriate and also provides an increased accuracy. Self designed dataset of sarcastic and non-sarcastic tweets is used for training of proposed model. Use of Recurrent Neural Network, Long Short Term Memory (LSTM) and Word Embeddings can make the sarcasm detection efficient and thereby making the statements from twitter easily classifiable.