基于提示增强原型网络的小样本多标签方面类别检测OA北大核心CSTPCD

Few-shot multi-label aspect category detection is a prominent area of research within fine-grained sentiment analysis.In methods based on prototypical networks,the lack of training data and the presence of irrelevant noise words severely impact the quality of prototype vector generation using attention mechanisms.Addressing this challenge,our study introduces the ProtPrompt model,an innovative prototype network enhanced through prompt learning.Firstly,we ali…查看全部>>