Text event extraction method based on multidirectional traversal and prompt learning

The invention discloses a text event extraction method based on multidirectional traversal and prompt learning. The method comprises the steps of collecting a corresponding event text, labeling entity information and trigger word information in the text, constructing an event data set, performing preprocessing, screening out data which does not meet requirements, and dividing the event data set into a training set, a verification set and a test set; the method comprises the following steps: constructing a prompt input template according to three sample traversal modes in combination with prompt learning, constructing an event argument extraction model based on a pre-training language model, training in combination with input, and finally realizing argument recognition and argument classification by utilizing the trained argument extraction model. The input information is trained and subjected to loss calculation according to three traversal modes, the problem that interaction among arguments is insufficient i