Prompt Learning for Event Detection

Prompt Learning is a new fine-tuning technique for pretrained language models (PLM), where prompt templates are designed and the model will make decisions based on the probabilities of the verbalizers of each class returned from the PLM.

In this project, we develop a sentence-level event detection system based on prompt learning. Given the sentence, the NLP system needs to detect all event types included in that sentence. We apply MAVEN which is the largest event detection dataset. It contains 168 event types in total.