A large model-based software compliance detection method and system

CN122240445APending Publication Date: 2026-06-19NO 30 INST OF CHINA ELECTRONIC TECH GRP CORP
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
NO 30 INST OF CHINA ELECTRONIC TECH GRP CORP
Filing Date
2026-02-05
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing software compliance testing methods are inefficient, difficult to adapt to diverse industry standards, and prone to misjudgments and omissions.

Method used

A software compliance detection method based on a large model is adopted. By preprocessing and semantically slicing the target compliance standard text, a ReAct-mode intelligent agent is constructed, including a reasoning module, an action module, and a memory module, to realize dynamic task planning and tool invocation, and generate a structured compliance detection report.

Benefits of technology

It achieves efficient and accurate software compliance testing with high coverage, reduces false alarms and omissions, adapts to various testing scenarios, and generates traceable testing reports.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122240445A_ABST
    Figure CN122240445A_ABST
Patent Text Reader

Abstract

This invention discloses a software compliance detection method and system based on a large model, involving software engineering and artificial intelligence. The method includes automated slicing and structured parsing of compliance standard text to generate a structured clause library. Subsequently, a compliance detection intelligent agent is constructed based on the ReAct pattern. This agent, through the collaborative work of its reasoning module, action module, and memory module, dynamically plans and executes a "knowledge retrieval-tool invocation" task sequence according to the structured clauses, achieving automated judgment of software compliance. Finally, a structured detection report is automatically generated based on the judgment results stored in the memory module. This application solves the problems of high maintenance cost and poor flexibility of existing rule-driven methods, as well as low efficiency and insufficient consistency of manual detection methods, achieving comprehensive, accurate, and traceable automated intelligent detection of software compliance.
Need to check novelty before this filing date? Find Prior Art