Large language model optimization method and storage medium

By constructing a knowledge graph in the power sector and optimizing the large language model using a dynamic weighted graph embedding algorithm, the problem of insufficient professional knowledge was solved, and the model's professional capabilities, accuracy, adaptability, and consistency in the field of power standards were improved.

CN122174875APending Publication Date: 2026-06-09SHANTOU POWER SUPPLY BUREAU OF GUANGDONG POWER GRID CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHANTOU POWER SUPPLY BUREAU OF GUANGDONG POWER GRID CO LTD
Filing Date
2026-01-30
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Large language models suffer from insufficient professional knowledge, limited depth of understanding, and lack of risk awareness in professional fields such as power, making it difficult to meet the professional requirements of power standard applications, and existing optimization methods are ineffective.

Method used

A knowledge graph is constructed based on industry-specific knowledge bases. Prompt templates are generated through graph embedding algorithms. Combined with dynamic weights and differentiated learning strategies, the basic large language model is optimized to generate a professional large language model. Knowledge applications are dynamically adjusted to adapt to standard updates.

Benefits of technology

This enhances the professional capabilities and application value of large language models in the field of power standards, improves the accuracy and adaptability of the models, and ensures consistency with the latest standards.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122174875A_ABST
    Figure CN122174875A_ABST
Patent Text Reader

Abstract

The application provides a large language model optimization method and a storage medium, and relates to the technical field of artificial intelligence. The method comprises the following steps: constructing a knowledge graph based on a knowledge base corresponding to an industry field, the knowledge graph comprising a plurality of nodes and connection relationships between the nodes, and the nodes representing knowledge points contained in the knowledge base; generating a prompt template corresponding to the knowledge graph based on a graph embedding algorithm; and optimizing a basic large language model based on the prompt template to obtain a professional large language model corresponding to the industry field. The application can improve the accuracy of the large language model in the application process.
Need to check novelty before this filing date? Find Prior Art