Unseen Target Stance Detection Based on Data Augmentation and Contrastive Learning
A stance detection model for unseen target based on data augmentation and contrastive learning is proposed for the stance detection task of unseen target.With the help of an unsupervised learning,the model distinguishes between non goal related and goal specific stance features by training the change of words related to the covered target and the covered target.To distinguish the irrelevant target/target specific stance feature types in the potential space to improve the quality of data embedding,a contrast hierarchical learning framework is adopted,which considers both augmentation signals and stance label information.On the public dataset,this model is compared with other models.The results indicate that the model has excellent performance.
万方数据
维普
