Sarcasm Detection in English-Hindi Code-Mixed Text Data

Sarcasm-also termed as verbal irony, when the speaker presents something in a mocking or humorous way such that he or she actually conveys the opposite of what they seem to say. This article has a detailed analysis of sarcasm detection in English-Hindi code-mixed texts based on the difficulties one faces while identifying sarcasm in linguistically complex data. Several machine learning and deep learning methods are adopted and compared, Long Short-Term Memory, Bidirectional Encoder Representations from Transformers, and Support Vector Machines are some of them. In addition, the complexity in the detection of sarcasm is further compounded by the nuance of code-mixed language because the former does not have any tradition that can be adapted to traditional text classification methods. Feature extraction and the subsequent decision-making processes are also factored with broader implications for sentiment analysis. The results of the models analysed underwent a comparative analysis, which showed that the model produced out of the combination of BERT and Gated Recurrent Unit (GRU) surpasses other models to be used for this particular task with its detection accuracy rate, said to be the best in determining sarcasm.