Supposititious Sarcasm Detection and Sentiment Analysis Coping Hindi Language in Social Networks Harnessing Zipf- Mandelbrot Probabilistic Optimisation and Perplexity Entropy Learning

Sarcasm is used to convey contempt via poking and ridiculing the other person. It is frequently used to make fun of other people by saying unpleasant things. The presence of informal language patterns, modified vocabulary, and speech-text utilized in online communities makes sarcasm detection a challenging task. Due to the occupancy of these terms in the conversation it is challenging to understand and identify sarcasm. Further, sarcasm detection in Hindi, a native Indian language is itself a challenging task to perform. This study proposes a novel method for sarcasm detection in Hindi comments taken from social networking sites using probabilistic non-linear optimization technique. The encapsulations of word embedding algorithms infused in an optimization problem is used for sarcasm detection. The Hindi datasets, i.e., news headlines and Twitter comments are utilized for sarcasm detection. The outcomes of proposed method is assessed using a variety of learning metrics such as accuracy, precision, recall, and F1 score surpassing the existing state-of-the-art methods. Performing ablation study and statistical validation tests within real-world social network confirms the efficacy and reliability of the proposed method. The proposed method gives an outstanding F1 scores of 0.9316% and 0.9359% on Dataset-1 and Dataset-2.