A structured data verification method and system based on multi-agent collaboration

By employing a multi-agent collaborative architecture, the problems of multi-dimensional field changes and cross-table logical reasoning in structured data verification are solved, achieving highly accurate and interpretable data verification, which is suitable for high-reliability scenarios such as finance and power.

CN122154733APending Publication Date: 2026-06-05STATE GRID SHANGHAI MUNICIPAL ELECTRIC POWER CO +1
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202610104429.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-01-26
Publication Date
2026-06-05

AI Technical Summary

Technical Problem

Existing fact-checking technologies struggle to handle multidimensional field changes, cross-table logical reasoning, and complex semantic expressions when processing structured data, resulting in poor interpretability, low accuracy, and a lack of dynamic response capabilities.

Method used

A multi-agent collaborative architecture is adopted, including agents for task parsing, indicator mapping, data retrieval, computation verification, and logic auditing. Through semantic embedding matching and logical consistency verification, it realizes automatic matching and verification of natural language and structured data.

Benefits of technology

It improves the accuracy and interpretability of structured data verification, generates traceable logical chains, and enhances the transparency and credibility of the system in high-reliability scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122154733A_ABST
    Figure CN122154733A_ABST
Patent Text Reader

Abstract

The application relates to a structured data verification method and system based on multi-agent cooperation, which comprises the following steps: acquiring power data or financial data, and performing structured conversion to form a standard data set; receiving natural language to be verified, matching the terms of the natural language with the standard data set, and performing structured representation on the natural language; performing problem analysis, index mapping, data retrieval, calculation verification and logic auditing processing on the structured representation of the natural language through multiple agents to obtain corresponding processing results; and uniformly integrating and analyzing the processing results of the agents to output corresponding data verification results. Compared with the prior art, the application significantly improves the transparency, reliability and practicability of the system in high-reliability scenarios such as financial analysis, power supervision and policy evaluation.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Financial text checking method and system

    CN116542252A

  • Automated fact checking using iterative knowledge base querying

    US12008000B2

  • Fact checking method and system

    US8185448B1