Sarcasm detection is the specific research area in natural language processing (NLP) that aims to identify irony containing utterance in the sentimental text from the user. A widely applied automatic detection model for sarcasm in NLP is a deep learning model. Sarcasm is a phenomenon with perlocutionary effects used to express intensified positive words. Intentional ambiguity in sarcastic sentences is difficult to capture using traditional deep learning approaches. Deep learning with an attention mechanism owns the potential to capture explicit and implicit contexts that are used to express sarcasm.
Psycholinguistic sources are the additional features that help analyze different cognitive and language level factors in sarcastic sentences. Attention-based sarcasm detection combined with psycholinguistic sources efficiently identifies the sarcasm with high accuracy.
• Sarcasm detection has gained popularity among Natural Language Processing (NLP) researchers, especially in the task of sentiment analysis.
• For efficacious sarcasm detection, attention neural networks need to be incorporated, conducive to capturing the complex relations in the sarcastic sentences.
• Sentiment comportment in sarcasm depends on the cognitive and linguistic features of the natural statement.
• Utilization of psycho-linguistic sources for discovering a set of potential features by assigning various psycho-linguistic categories to the sarcastic statement.
• Sarcasm detection using an attention neural model captures the complex expression of sarcastic sentences to learn sentiment inconsistencies.
• Psycho-linguistic sources help attention guided sarcasm detection as additional knowledge to identify the sarcastic term, regrading psychological and linguistic aspects.