Sarcasm Detection in News Headlines Using Deep Learning

A communication technique known as sarcasm is used to express an emotion that is opposite or distinct from the literal meaning of the words being used, typically for satire, criticism, or amusement. Sarcasm detection is the process of identifying language that purposefully conveys the opposite meaning of what is being said. Today's sarcasm detection techniques employ a range of deep learning (DL) and machine learning (ML) algorithms, including logistic regression, random forests, decision trees, long short-term memories (LSTM), convolution neural networks (CNN) etc. However, these methods have limitations in accurately detecting sarcasm, particularly when dealing with large datasets. In this paper, we aim to provide a better method for sarcasm identification in news headlines that combines LSTM and CNN algorithms enhancing sarcasm detection's accuracy and here we managed to achieve nearly greater than 97% of accuracy by leveraging the strengths of both algorithms. Using a dataset of news headlines, we have also compared our hybrid model to other models to see how effective our proposed approach is more accurate.