Sarcasm Detection Using Dimension Glove Embedding
R - Sanchana., V. Brindha Devi, C. Valarmathi, P. Mercy, L. Meenakshi · 2022 1st International Conference on Computational Science and Technology (ICCST) · 2022
Sentiment analysis is one of the distinguished premises of data prospecting. The distinguished mining field deals with the identification and analysis of sentiment within the text. Sentiment analysis is the process of decisive approach to actuate whether the contradiction in a particular content is explicit, negative, or unbiased. The main challenges in sentiment analysis are detecting sarcasm and irony in a text. Sarcasm detection in sentiment analysis is a challenging task to fulfill without having a complete perceptive of the vocabulary of the text. There are different types of sentiment. One such type of sentiment is sarcasm where individuals offer their viewpoint in a contrary manner. While speaking people often use tonal accent and genuflection clues like reverberating eyes and hand movements to proclaim sarcasm. When we try to deal with sarcasm in textual content, gesture and tonal stress are missing. The gesture and tonal stress which are missing in textual content make sarcasm detection in a text a very troublesome task. Therefore the system has to be trained. The system has to be trained in such a way for determining whetherthe peculiar tweet is sarcastic or non-sarcastic. The main objective is to identify the best glove word dimension embedding which produces better accuracy which in turn can segregate the tweet as sarcastic or non-sarcastic. The tweets are pre-processed. The tweets are then passed as input to the different Glove embedding models. The glove model creates a word co-occurrence in a massive text corpus to generate a large dimension of words. The turnout of the glove model is given as instruction to the different deep learning techniques like LSTM. The accuracy obtained by the LSTM model using 50, 100, 200, and 300 dimension Word Embeddings are 79%, 81%, 78%, and 83%. The 300 dimensions have achieved better accuracy when compared to other dimensions.