Comparative Study of Sarcasm Detection Using Glove + LSTM Technique in Machine Learning

Natural language processing (NLP) finds it difficult to detect sarcasm in text since it depends on irony, context, and frequently contradicting cues. Applications like sentiment analysis, content moderation, and fake news detection all depend on an understanding of sarcasm. We compare sarcasm detection methods in this study utilizing two popular machine learning(ML) techniques: GloVe embeddings and LSTM (Long Short-Term Memory) networks. GloVe + LSTM's performance in detecting sarcasm in text is compared to that of other deep learning architectures (convolutional neural networks(CNN), vanilla RNNs) and conventional machine learning(ML) models (Support Vector Machine(SVM), Naïve-Bayes(NB)). Our findings show that, in terms of accuracy, precision, recall, and F1 score, the GloVe + LSTM model performs better than conventional models. This study also highlights the effectiveness of GloVe embeddings for word representation and LSTM networks for capturing long-term dependencies in text, offering insights into the advantages and disadvantages of different sarcasm detection techniques.