A verifiable control method and system for employee subject authorization verification and authorization change

By parsing natural language commands using a neural network model and combining it with cross-index mapping tables and idempotent authorization changes, the accuracy and consistency issues of authorization verification and changes under large-scale authorization object sets are solved, achieving low-cost, high-efficiency, and verifiable control.

CN121544219BActive Publication Date: 2026-05-05ZHONGJIAN GROUP GONGXIN SECURITY TECHNOLOGY CO LTD
2 Cites 0 Cited by

Patent Information

Application Number
CN202610084803.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-01-22
Publication Date
2026-05-05
Estimated Expiration
2046-01-22

AI Technical Summary

Technical Problem

With large-scale sets of authorized objects, existing technologies suffer from problems such as out-of-bounds output, format drift, semantic crosstalk, and difficulty in recovering from erroneous operations. Especially under lightweight deployment conditions, the accuracy and consistency of authorization verification and changes are difficult to guarantee.

Method used

A neural network model is used to parse the natural language commands of the employee subject. Through cross-index mapping tables and idempotent authorization changes, combined with compensating transaction logs and hard check gates, the verifiability and consistency of authorization verification and changes are ensured.

Benefits of technology

With low-cost and lightweight deployment, it achieves efficient and accurate authorization verification and modification, and has verifiability, controllability and recoverability, avoiding out-of-bounds writes and maintaining the consistency and traceability of large-scale task sets.

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

Abstract

This invention discloses a verifiable control method and system for employee subject authorization verification and authorization changes. The method segments the natural language authorization task and establishes session isolation for each subtask; in the first session, a neural network model is invoked to output a fixed-format set of employee subject names, action enumerations, and task phrase sets; a candidate task subset is generated based on cross-indexing, and in the second session, a neural network model is invoked to perform constrained semantic alignment, allowing the output of candidate task codes or newly created task names; the execution layer rejects out-of-bounds output through a hard check gate; a compensation transaction log is generated before writing and a snapshot and fingerprint value of the affected domain file before the change are saved; after writing, the fingerprint value is recorded and a commit flag is written; in case of abnormal restart, rollback recovery can be performed based on the transaction status; the authorization cache is maintained or rebuilt according to the incremental and differential sets, providing reversal or compensation transactions based on fingerprint value conflict detection.
Need to check novelty before this filing date? Find Prior Art