Electronic text event extraction method and device, equipment and storage medium

Abstract

The invention discloses an electronic text event extraction method and device, equipment and a storage medium. The method comprises the following steps: preprocessing an original data set, and dividing the preprocessed original data set into a training set, a verification set and a test set; according to the training set and the verification set, training a context-integrated multi-task joint electronic text event extraction network, and generating a multi-task joint electronic text event extraction model; predicting the test set according to the multi-task joint electronic text event extraction model, extracting a test result of each word in the test set, realizing introduction of context-related text information, taking text representation related to extracted sentences in the context as auxiliary representation, helping the model to distinguish events and roles which are easy to confuse, and improving user experience. Meanwhile, the context representation serves as a feature bridge among the tasks, interaction among the tasks is facilitated, the extraction accuracy is improved, and interaction among the subtasks is enhanced through the context representation.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, and in particular to an electronic text event extraction method, device, computer equipment and computer-readable storage medium. Background technique [0002] With the continuous development of science and technology and the popularization of the Internet, the Internet has become an indispensable tool in people's life and work. Countless information is presented in front of us in an explosive manner, filling our current life without any effective filtering. Efficient access to effective information has encountered unprecedented challenges. Event extraction (event extraction EE) performs fast and accurate information mining from a large amount of unstructured text, and extracts the information required by users, which has important practical significance. The event extraction task can be divided into three subtasks, including entity recognition task, trigger word recogn...

AI Technical Summary

Problems solved by technology

[0006] The main purpose of this application is to provide an electronic text event extraction method, device, computer equipment, and computer-readable storage medium, aiming at solving the problem that most joint models at the present stage extract entity type information as known information. In the joint extraction process, only the trigger word extraction task and the argument role prediction task are combined, making the model unable to obtain the inherent deep semantic connection between the entity task and the other two tasks.

Images