Enhanced Sarcasm Detection using Grey Wolf Optimizer with Deep Learning on Social Media

A linguistic expression, Sarcasm generally conveys the opposite meaning of what had been said in words, which makes it a compplex process for the machines to find out the real meaning. It is mainly depends on the context, making it hard for computation task and is largely defined by the inflexion with which it is spoken, with suggestion of irony. Furthermore, sarcasm express the negative sentiments using positive word, enabling it to confuse sentiment analysis (SA) model. Sarcasm detection is a crucial challenge in natural language processing (NLP), which is required for good understanding to serve as an interface for mutual interaction between humans and machines. This article develops an Enhanced Sarcasm Detection using Grey Wolf Optimizer with Deep Learning (ESD-GWODL) technique on social media. The major intention of the ESD-GWODL method is to detect and classify sarcasm on social media. To accomplish this, the ESD-GWODL technique applies data pre-processing to transform the input data into relevant format. Besides, TF-IDF model can be used to extract word embedding. To detect and categorize sarcastic texts, the ESD-GWODL technique applies long short-term memory (LSTM) model. The ESD-GWODL technique utilizes GWO algorithm for hyperparameter tuning purposes in order to boost the sarcastic text detection. The experimental evaluation of the ESD-GWODL technique is tested using sarcasm dataset. The obtained values infer that the ESD-GWODL technique has resulted in boosted detection results.