Emotion Recognition Based on Hierarchical Graph Convoluation Networks
The fine-grained emotion recognition involves more emotion categories and can accurately discriminate e-motions that individuals experience and express in their daily lives.To consider deep sentiment knowledge like emo-tional hierarchy relationships,this paper proposes a Hierarchical Graph Convolution Networks based Emotion Rec-ognition(HGCN-EC)model.The HGCN-EC model consists of three modules:a semantic information module,an emotion hierarchy knowledge module,and a knowledge fusion module.The semantic information module extracts the semantic features of texts.The emotion hierarchy knowledge module organizes fine-grained emotions into a tree-like hierarchy and calculates the transition probabilities between emotions as hierarchy knowledge based on Bayesian statistical inference.The knowledge fusion module uses graph convolutional network to fuse hierarchy knowledge of emotions with text semantic information for emotion prediction.Experiments on the GoEmotions dataset show that the proposed method outperforms baseline methods.
万方数据
维普
