Man-machine question and answer interaction system for financial examination and approval platform

By introducing a human-computer question-and-answer interaction system into the financial approval system, and utilizing the BERT model and named entity recognition technology, a natural language-driven dynamic approval path was realized. This solved the problems of rigid interaction and high rule coupling in the financial approval system, and improved efficiency and exception handling capabilities.

CN120910210APending Publication Date: 2025-11-07YAHANG SHARED FINANCE & TAXATION (ZHONGSHAN) CO LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202511042807.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-28
Publication Date
2025-11-07

AI Technical Summary

Technical Problem

The existing financial approval system suffers from rigid interaction, high rule coupling, and inefficient exception handling, resulting in complex reimbursement processes, difficulty in changing rules, and long processing times for exceptions.

Method used

The financial approval platform, based on natural language processing, includes a user interaction module, a semantic parsing engine, a rule matching engine, an approval decision engine, and a feedback execution module. It uses the BERT model and named entity recognition technology to achieve dynamic approval paths and real-time rule collaboration, and supports natural language queries and automated anomaly handling.

Benefits of technology

It improved financial approval efficiency by 40%, reduced rule maintenance costs by 60%, increased exception handling speed by 5 times, and achieved natural language parsing and dynamic policy adaptation.

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

Abstract

The invention discloses a man-machine question and answer interaction system for a financial examination and approval platform. The man-machine question and answer interaction system comprises a user interaction module for receiving natural language query and outputting a structured examination and approval instruction; the semantic analysis engine is connected with the user interaction module and decomposes natural language query into an intention tag and an entity parameter; the rule matching engine is connected with the semantic analysis engine, matches a preset financial examination and approval rule base based on the intention label, and injects entity parameters to generate a dynamic examination and approval path; the examination and approval decision engine is connected with the rule matching engine, executes logic judgment in the dynamic examination and approval path and generates an examination and approval decision result; and the feedback execution module is connected with the examination and approval decision engine, converts the decision result into natural language feedback and triggers the financial system to execute operation. Due to the adoption of the technical scheme, compared with the existing approval system, the efficiency is improved by at least 40%; the substitute document filling can be analyzed through a natural language; the rule maintenance cost is reduced by at least 60%; the dynamic path generation supports real-time updating of policies; the exception handling response speed is increased by more than 5 times; and a rechecking mechanism can be automatically triggered.
Need to check novelty before this filing date? Find Prior Art