Sarcasm Detection in Social Media Text Using GloVe Word Embeddings in Machine Learning

This paper investigates the use of Global Vectors for Word Representation (GloVe) in detecting sarcasm within textual data. By integrating GloVe embeddings with machine learning models, Enhancing sarcasm detection accuracy is our goal. Our approach employs a combination of GloVe embeddings and word2vec technique to capture both semantic and contextual nuances. For humans, sarcasm has always been a difficult idea. Sarcasm recognition due to interesting linguistic properties Natural language processing (NLP) has attracted increasing attention from the field of research in recent years. Even machines that have little understanding of what constitutes a sarcastic remark nonetheless find it challenging to forecast sarcasm in a text. The study contrasts various machine learning techniques using GloVe and word2vec representations, including logistic regression, random forests, and deep learning models. The findings show that word2vec and GloVe both offer useful features for sarcasm detection, with word2vec outperforming GloVe by a little margin in terms of accuracy and precision. We further investigate the function of sentence-level features and contextual information, indicating that adding linguistic signals in addition to word embeddings improves model performance. Studies indicate that the problem of sarcasm identification in social media material may be resolved with the use of embedding-based methods when combined with conventional machine learning(ML) techniques.