Entity relation joint extraction method and device for discontinuous entity

Through the entity relationship extraction model, hollow convolution and joint classifier are used to process the character combinations of non-continuous entities, which solves the problem of inaccurate non-continuous entity relationship extraction in the existing technology and achieves more efficient entity relationship extraction.

CN118551764BActive Publication Date: 2025-10-17INST OF AUTOMATION CHINESE ACAD OF SCI
1 Cites 0 Cited by

Patent Information

Application Number
CN202411028372.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2025-10-17
Estimated Expiration
2044-07-30

AI Technical Summary

Technical Problem

Existing technologies are difficult to effectively handle the relationships between non-continuous entities, resulting in inaccurate entity relationship extraction.

Method used

An entity relationship extraction model is adopted, and the hollow convolution layer and joint classifier are used to predict the labels of character combinations. The context information, distance information and relative position relationship information of the characters are combined. The model is trained through the table filling method, special labels are designed for data annotation, and the cross entropy loss is calculated for model training.

Benefits of technology

The adaptability and accuracy of the model in processing non-continuous entity relationships are improved, and the extraction effect in complex scenarios is enhanced.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application provides an entity relation joint extraction method and device for non-continuous entities. The method belongs to the technical field of information extraction and natural language processing, and comprises the following steps: predicting the relation between every two characters in a to-be-processed text by using an entity relation extraction model; for training of the entity relation extraction model, firstly, storing every two characters in a training text as a character combination in a table, and marking the relation between the characters in the table; predicting the relation label between the characters of the character combination by using a hollow convolution layer and a joint classifier in the model; and training the model by calculating the loss based on the real label information and the predicted label information. The entity relation joint extraction method and device for non-continuous entities provided by the application realize joint extraction of the relation of non-continuous entities by using the hollow convolution and the joint classifier to predict the label of the table formed by the character combination, and improve the adaptability of the model in complex scenes.
Need to check novelty before this filing date? Find Prior Art