Text Emotion Recognition Using GRU Neural Network with Attention Mechanism and Emoticon Emotions

Abstract

In this study, we propose an emotion identify model called SEER, in this model, we first combined a Bi-directional Gate Recurrent Unit (Bi-GRU) network and attention mechanism to capture the emotion vectors for the aspect of input words, and second, we statistically analyze the emoticon that appears in our data set to obtain the emoticon distribution, then, use the emoticon distribution to enhance the emotion vectors. The experiment proved that combine with Attention Mechanism and Emoticon Distribution is an effective way to improve the accuracy of emotion recognition. Compared with other deep learning methods, machine learning methods, and other methods, the experimental results show that the method we posed in this paper has achieved the highest accuracy in emotion recognition.