Negative Emotion Recognition Algorithm of Network Catchwords Based on Language Feature Dimension

Abstract

The traditional negative emotion recognition algorithm has a limited language feature dimension, which leads to the inaccuracy of negative emotion recognition. In order to improve the identification and analysis of emotion in network buzzwords, the back propagation of error (BP) and the restricted Boltzmann machine (RBM) algorithms are adopted to effectively solve the problem of insufficient data for emotion analysis in different contexts. First, a method is proposed to identify negative emotions, and a deep neural network (DNN) model is constructed. Then, experiments were carried out, which used manually labeled data sets and divided them into different emotion categories, and which compared the BP algorithm, Gaussian Mixture Model (GMM) and Hidden Markov Model (HMM) for negative emotion recognition of online buzzwords. The experimental results show that the DNN model performs well in the recognition of anger, sadness, fear and disgust, with the accuracy reaching 93.56%, 93.58%, 89.84% and 88.53% respectively, which is obviously superior to the other three methods. The designed DNN model has a potential application prospect in the negative emotion recognition of online buzzwords, which can be further popularized in the future.