A method and device for large model-based structured parameter output for financial scenarios

By combining document type recognition, strategy planning agents, and multi-path extraction agents with financial business logic constraint graphs, the accuracy and consistency issues of structured parameter extraction in financial scenarios are solved, achieving efficient and low-cost financial document parameter extraction.

CN122045299BActive Publication Date: 2026-07-03CSC FINANCIAL CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
CSC FINANCIAL CO LTD
Filing Date
2026-04-15
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Existing technologies struggle to accurately extract structured parameters in financial scenarios, especially when dealing with complex unstructured documents, exhibiting poor generalization ability. Furthermore, large models are costly to use and have low confidence levels, leading to inconsistent and erroneous extraction results.

Method used

By combining document type recognition, strategy planning agents, and multi-path extraction agents with financial business logic constraint graphs, accurate extraction and consistency verification of financial documents are achieved. This includes a hybrid arrangement of regular expression extraction, enhanced retrieval generation, and large model inference to resolve conflicts.

Benefits of technology

It improves the accuracy and robustness of structured parameter extraction, reduces computational costs, and solves the challenges of semantic understanding and logical reasoning under complex layouts.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122045299B_ABST
    Figure CN122045299B_ABST
Patent Text Reader

Abstract

The application provides a structured parameter output method and device based on a large model for a financial scenario, and relates to the technical field of computers. The method provides accurate preconditions for subsequent processing through document type identification; optimizes the modeling of parameter extraction planning problems through a strategy planning agent, adaptively plans the extraction path according to the parameter and extraction method adaptation degree, and realizes the mixed arrangement of various technical means; through a multi-path extraction agent, the accuracy of regular expressions, the knowledge retrieval ability of search enhancement generation technology and the logical reasoning ability of a large model are comprehensively utilized to realize deep understanding of financial business parameters; and the consistency of the preliminary structured parameter data is checked, and conflict resolution processing is performed to obtain the final structured parameter result and output. The scheme reduces the computing cost while improving the accuracy and robustness of the extraction result, and solves the semantic understanding and logical reasoning problem under a complex format.
Need to check novelty before this filing date? Find Prior Art