Sarcasm Detection of Facebook's Posts Using Machine Learning Models

The coronavirus disease 2019 (COVID-19) has brought massive challenges to the world, altering people's lives. We conducted a study on people's mental states during Taiwan's National Epidemic Level 3 Alert in 2021. On May 22, 2021, during a regular press conference held by the Taiwan Centers for Disease Control (CDC), the Minister of Health and Welfare introduced the term 'Retrospective Adjustment', which left the entire population in shock. Our approach consists of a two-stage task. Constructing the model is the first stage and detecting sarcasm is the second. First, we integrated TCNN and BiLSTM models. Second, we focused on misclassified by the model and performed feature engineering based on these misclassified data. After constructing features, we performed well using machine learning models. Finally, the experimental results show that our approach performs well in detecting sarcasm using linguistic and lexical-based features. In the second stage, the LSTM model detects sarcasm, achieving a performance of 0.73. We integrate the results of two stages to adjust accuracy. By improving the accuracy of the misclassified data to 0.6, the overall accuracy for negative posts has increased to 0.76.