Multi-label Hierarchical Sentiment Text Classification Method Based on Contrastive Learning
When dealing with multi-label emotion classification tasks,traditional emotion classification meth-ods are often difficult to capture the dependency between labels effectively,which leads to poor classification performance.In order to solve the difficult problem of complex relationships between labels in multi-label emo-tion classification task,a multi-label hierarchical emotion text classification method based on contrast learning is proposed.By introducing GoEmotions labeling system,hierarchical labeling representation based on graph neural network,combined with comparative learning strategy,the aim is to optimize the representation space of the model and improve the classification accuracy.Experiments were conducted on GoEmotions and ISEAR datasets to verify the effectiveness of the proposed method.The experimental results show that the proposed method performs significantly better than the existing baseline method in multi-label emotion classification tasks.
万方数据
维普
