Bilateral relationship knowledge base construction method and device based on multi-task deep learning

By adopting a multi-task deep learning model in the construction of the national bilateral relationship knowledge base, joint training and information interaction of each task are achieved, the problem of waste of resources and insufficient collaborative optimization caused by independent task training in the existing technology is solved, and the overall quality and coverage of the knowledge base are improved.

CN119721211AActive Publication Date: 2025-03-28EVALUATION & DEMONSTRATION RES CENT OF THE CHINESE PEOPLES LIBERATION ARMY ACAD OF MILITARY SCI
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411811582.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-10
Publication Date
2025-03-28
Estimated Expiration
2044-12-10

AI Technical Summary

Technical Problem

When the existing technology is constructing a national bilateral relationship knowledge base, independent training of each task leads to repeated calculations and waste of resources, lack of collaborative optimization between tasks, and complex deployment and maintenance.

Method used

Using a multi-task deep learning method, a multi-task deep learning model is built, including an input layer, an embedding layer, a shared encoding layer, a specific task layer, an output layer and a result processing layer. Through the design of shared encoding layer and specific task layer, joint training and information interaction of each task are realized.

Benefits of technology

It realizes joint learning of multi-tasks, reduces duplicate calculations, improves the collaborative optimization effect between tasks, simplifies data annotation and system deployment, and improves the overall quality and coverage of the bilateral relationship knowledge base.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119721211A_ABST
    Figure CN119721211A_ABST
Patent Text Reader

Abstract

The invention relates to a bilateral relation knowledge base construction method and device based on multi-task deep learning. According to the method, a multi-task deep learning model is shared by four sub-tasks for constructing a national bilateral relationship knowledge base, and four tasks are jointly learned in the same model, so that entity extraction, abstract generation, news classification and sentiment analysis can be simultaneously performed on news texts; and outputting complete information including bilateral countries, time, news abstracts, news types and emotion types, and finally efficiently constructing a large-scale bilateral relationship knowledge base covering multiple levels of the national bilateral relationship. By adopting the method, the problems of independent training of each sub-task, lack of collaborative optimization, data annotation, high model deployment cost and the like in existing bilateral relationship knowledge base construction are effectively solved, efficient construction of news text understanding and a national bilateral relationship database is realized, and the method has remarkable technical advantages and application value.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • News emotion analysis method based on multi-model fusion

    CN112507723A

  • Internet news analysis system and method based on big data

    CN118093979A

  • Domain large model multi-modal knowledge base construction method based on feature representation

    CN118779469A

  • Semantic similarity-based entity relation extraction method and apparatus, device and medium

    WO2021121198A1