A smart data analysis method and system integrating Chat BI and Headless BI

By constructing a unified semantic data model and knowledge graph, combined with a large language model, the technical barriers of Chat BI and Headless BI have been overcome. This has enabled the transformation from natural language interaction to standardized data querying, improved the accuracy of SQL generation and data querying, and made it suitable for enterprise-wide data analysis.

CN122132495APending Publication Date: 2026-06-02NARI INFORMATION & COMM TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
NARI INFORMATION & COMM TECH
Filing Date
2026-05-07
Publication Date
2026-06-02

AI Technical Summary

Technical Problem

Existing Chat BI and Headless BI technologies suffer from semantic bias and high operational barriers, respectively, making it difficult to balance interactive convenience and analytical professionalism in enterprise data analysis, resulting in a trade-off between data query efficiency and accuracy.

Method used

By constructing a unified semantic data model and knowledge graph, combined with a large language model (LLM), the transformation from natural language interaction to standardized data query is realized, including semantic verification and permission injection, generating accurate semantic SQL and executing it, and supporting multimodal display.

Benefits of technology

It combines the convenience of natural language interaction with professional and standardized data analysis, improves the accuracy of SQL generation and data query, adapts to multi-source heterogeneous databases, reduces the risk of large model illusion, and is suitable for enterprise-wide data analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122132495A_ABST
    Figure CN122132495A_ABST
Patent Text Reader

Abstract

This invention discloses an intelligent data analysis method and system integrating Chat BI and Headless BI; belonging to the field of artificial intelligence and data analysis technology, the operation steps are as follows: Building a unified semantic data model based on the semantic layer of Headless BI, constructing a knowledge graph of data table associations; receiving natural language query requests and completing intent parsing through a large model; calling the semantic layer of Headless BI to generate standardized semantic SQL; after verifying and injecting permissions into the semantic SQL, converting it into physical SQL and executing it; displaying the query results and generating intelligent analysis summaries through the Chat BI multimodal module. The system includes a large model parsing module, a semantic layer processing module, an SQL conversion and verification module, a multimodal display module, a Chat BI interaction module, and a data storage module. This invention achieves deep integration of Chat BI and Headless BI, balancing convenient interaction with professional analysis, improving SQL generation accuracy, reducing the risk of large model illusion, efficiently supporting complex enterprise-level business queries, and adapting to multi-source heterogeneous databases.
Need to check novelty before this filing date? Find Prior Art