Combining Deep Learning and Contextual Handcrafted Features for Sarcasm Identification in Tweets

Abstract

Sarcasm identification in tweets poses a unique challenge due to the informal nature of the text and the subtlety of sarcasm. In this paper, we propose a novel approach that combines deep learning techniques with contextual handcrafted features for effective sarcasm identification in tweets. Our methodology involves preprocessing the tweet data, extracting both deep learning representations and handcrafted features, and combining them to train a hybrid LSTM-CNN model. We present a comprehensive evaluation of the proposed approach using a real-world dataset, showcasing its scalability and performance. Through extensive experimentation, we demonstrate that our model achieves state-of-the-art results in terms of both efficiency and accuracy, effectively capturing the nuances of sarcasm in tweets.