A comprehensive review of automated sarcasm detection techniques on Twitter

Automated sarcasm recognition is a valuable tool for analysing user sentiment on social media platforms like Twitter and Facebook. Sarcasm, a form of criticism, involves expressing thoughts contrary to the intended meaning, often as a means of insult or humour. Failing to recognise sarcasm can limit social and professional interactions. Our approach for sarcasm recognition utilises user embedding fitting, requiring only the user's previous text posts without complex feature engineering or simultaneous data harvesting. In comparison to a state-of-the-art strategy relying on an extensive attribute collection, our model demonstrates superior performance. We evaluated various classifiers, including ARABERT, SVM, LSTMs, CNN, and Pattern-based methods, on Twitter sarcasm detection. CNN achieved the highest accuracy at 95%, followed by LSTMs at 91.60%.