Performance Analysis of Clan Updated Grey Wolf Optimization (CU-GWO) based Deep Ensemble Technique for Identifying Sarcasm from Social Media Data
Geeta Abakash Sahu, Manoj Hudnurkar · 2023
Sarcasm is often considered as a prevalent linguistic element in online content that conveys strongly held subjectivity and viewpoints. The detection of sarcasm is a significant consideration over the dominant NLP appliances as sentiment analysis, which is utilizing for advertising and opinion mining is a precious tool for figuring out the people sentiments and attitudes. The proposed approach describes automatic detection of sarcasm as an ease classification of text challenges. This paper aims to develop a sarcasm detection model that categorizes words into either sarcastic or non-sarcastic expressions. Initially, the text input involves processing, here the stop word removal and tokenization process are carried out. Subsequently, various features like information gain, chisquare, mutual information, and symmetrical uncertaintybased features are extracted from the preprocessed data. The curse of dimensionality is a crucial issue, thereby the selection of optimal features is conducted by employing hybrid optimization model called Clan Updated Grey Wolf Optimization (CU-GWO). The optimal features are utilized for sarcasm detection by using ensemble technique. The ensemble techniques involve the classifiers such as NN, SVM, RF and DCNN. The final output of DCNN indicates whether the sarcasm is present. The effectiveness of the proposed approach demonstrated through diverse metrices.