Proposing System to Recognize Emotions in Public Network Using Phobert Deep Learning Model

People can receive information faster especially in the 4.0 revolution with the continuous development of revolutions. The information can affect our emotional, psychological and spiritual well-being, especially in the recent high school graduation exam across the country. Therefore, we propose to build a user emotion analysis system in a public network with the PhoBert training model in the paper. Besides, we built our dataset aggregated from social networks, articles, blogs, etc. We next use the PhoBert model to solve the processing data problems. The simulation results have shown an accuracy of 86.5% on the training and 81.32% on the validation dataset with a training time of 3 hours (about 180 minutes). The results also show that we can build a warning system to avoid health and psychological effects with great emotion.