Enhancing Low-resource Prompt-based Relation Representation with Multi-view Decoupling Learning

The author highlights the importance of improving low-resource prompt-based relation representations through multi-view decoupling learning. By introducing MVRE, the method aims to optimize relation inference by decoupling relations into multiple perspectives and leveraging Global-Local Loss and Dynamic Initialization techniques.