Method for converting natural language of a full link into structured query language

By combining knowledge graphs and large language models, the shortcomings of NL2SQL technology in understanding complex queries and adapting to different database structures are addressed, achieving efficient and accurate natural language to SQL query conversion, adapting to various scenarios and databases.

CN118227653BActive Publication Date: 2026-07-24BEIJING INST OF TECH
2 Cites 0 Cited by

Patent Information

Application Number
CN202410434738.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-11
Publication Date
2026-07-24
Estimated Expiration
2044-04-11

AI Technical Summary

Technical Problem

Existing NL2SQL technology suffers from insufficient depth of understanding, weak generalization ability, and inadequate flexibility and scalability when dealing with language diversity, complex queries, contextual understanding, and database structure dependencies.

Method used

By combining knowledge graphs and open-source large language models, and through pre-training models, building corpora for specific application scenarios, optimizing model parameters, and integrating the large language model into the user interface, we can achieve efficient conversion of natural language to SQL queries.

Benefits of technology

It improves the accuracy and flexibility of understanding natural language, enhances the system's scalability and generalization capabilities, lowers the technical threshold for users, and improves query efficiency and accuracy.

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

Abstract

The application discloses a kind of full-link natural language conversion to structured query language method, belong to data processing field, including the following steps: S1, pre-training open source large language model is used to existing corpus and prompt engineering;S2, based on knowledge graph, construct NL2SQL corpus for specific application scene, obtain training set and test set;S3, obtain best open source large language model;S4, input natural language to best open source large language model, query and generate SQL statement.The application adopts the above-mentioned full-link natural language conversion to structured query language method, combines the structured knowledge of knowledge graph and the powerful natural language processing capability of open source large language model, can better understand the entity in query and their relationship, so as to accurately capture the query intention of user, even when facing complex or fuzzy query can also maintain high understanding accuracy.
Need to check novelty before this filing date? Find Prior Art