SARCASM DETECTION USING MACHINE LEARNING BY TWITTER DATA

COMPUTER SCIENCE ENGINEERING

Complex linguistic term sarcasm is frequently used in social media and e-commerce websites. In Natural Language Processing applications like sentiment analysis and opinion mining, failure to recognise sarcastic utterances will confound classification algorithms and produce erroneous results. Numerous studies on the detection of sarcasm have used various learning methods. The majority of these learning methods, though,have always placed their primary attention solely on the expressed ideas, disregarding the context. AsAs a result, they missed the sarcastic expression's semantics and background information. Secondly,A word embedding learning technique is widely used in NLP deep learning approaches as a typical method for convolutional feature vector representation, which does not take into account the polarity of the words emotional connotations.To solve the challenges noted above, this work suggests a context-based feature technique for sarcasm identification using the deep learning model, BERT model, and traditional machine learning. For the categorization, two Twitter and Internet Argument Corpus, version two (IAC-v2) benchmark datasets were used.the three learning models are employed. The initial model makes use of deep learning and embedding-based representation.recurrent neural network with a bidirectional long short term memory (Bi-LSTM) (RNN),the development of word embedding and context using Global Vector representation (GloVe)learning. The second model is constructed using a pre-trained Bidirectional Encoder representation and is based on Transformer.both Transformer (BERT). The third model, by contrast, is based on the feature fusion of the BERT feature.a feature with sentiment-related, syntactic, and GloVe embedding

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.