Sarcasm detection identifies irony and incongruous containing utterances in the sentimental text from the user. The significant goal of sarcasm detection is to predict whether a sentence is sarcastic or non-sarcastic by the understanding of the user-s opinions and sentiment. Sarcasm is a specific anomaly with perlocutionary effects and sentiment transition.

  Recognizing such emotional transition is difficult while using classic sarcasm detection methods due to changing sentiment polarity in sarcastic sentences. Contextual embedding architecture in a deep learning model possesses the potential to capture the sentiment polarity of the words in the sarcastic expression. Contextual embedding is the feature vector representation of text based on the context that assists sarcasm detection to recognize the emotion transition in the ironic utterances efficiently.

  • Sarcasm detection facilitates the understanding of people’s true sentiments and opinions and benefits many areas of interest in natural language processing tasks.

  • Even though sarcasm detection is a complicated task, it is highly dependent on context, prior knowledge, and the tone in the sequence of text or audio.

  • Recognizing the emotion transition is beneficial for sarcasm detection as it explores more fine-grained categories of emotions.

  • Contextual embedding techniques with deep learning models extract appropriate features of sentimental polarity variations from the text or audio content.

  • In sarcasm detection, the emotional changes are identified effectively using the contextual embedding model.