LLM Code Violation Detection for Automated Deviation Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing code and deviation management systems require significant manual effort and are prone to errors, inefficiencies, and delays due to the lack of automated solutions for identifying coding rule violations and managing deviations.
Innovation Solution
A system utilizing a large language model (LLM) to analyze source code for coding rule violations, provide feedback, and manage deviations through intelligent code modification and report creation, enabling real-time interaction with users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual code analysis and deviation management procedures are used, then users can review and manage coding rule violations, but significant time and effort are required and errors are prone to occur
Solution Approach 1:
The system enables self-service by automatically detecting coding rule violations, generating deviation reports, and managing the entire deviation process without requiring manual intervention for routine tasks. The system serves itself by autonomously analyzing code, comparing against coding rules, and initiating deviation workflows.
Solution Approach 2:
The patent replaces manual mechanical processes with an automated system. Instead of users manually reviewing code and creating deviation reports, the system uses automated detection mechanisms, AI-based analysis, and electronic workflows to substitute human actions, thereby reducing time loss while maintaining reliability.
2Reliability
If manual deviation management procedures are used, then users can review and approve deviations, but the process is time-consuming and inefficient
Solution Approach 1:
The system implements feedback mechanisms where deviation reports are automatically generated and presented to users for review, and user approvals or modifications are automatically processed and reflected in the system state. This automated feedback loop maintains accuracy while significantly improving processing efficiency.
Solution Approach 2:
The system introduces an intermediary automated system between code analysis and deviation approval processes. This intermediary handles routine tasks such as violation detection, report generation, and workflow initiation, allowing human users to focus only on complex decision-making aspects.
3Extent of automation
If automated code analysis is implemented, then coding rule violations can be detected automatically, but the system needs to handle complex violations requiring user feedback
Solution Approach 1:
The system segments the violation handling process into distinct stages: automatic detection phase, report generation phase, user review phase, and deviation management phase. This segmentation allows automation to handle straightforward violations while directing complex cases to users for feedback, managing overall system complexity.
Solution Approach 2:
The system dynamically adjusts its behavior based on violation complexity. Simple violations are automatically processed, while complex violations trigger user feedback mechanisms. The system adapts its automation level in real-time based on the nature of detected violations, balancing automation extent with complexity management.
Data Source
AI summary
Provided are embodiments for code and deviation management. A method implementing a large language model (LLM) may be provided. The method may include: receiving, from a user, a source code; determining whether or not the source code violates at least one coding rule; based on determining that the source code violates the at least one coding rule, presenting information associated with the violation to the user; receiving, from the user, a user feedback associated with the presented information; determining, based on the user feedback, whether or not at least one rule deviation is to be implemented; based on determining that the at least one rule deviation is to be implemented, performing at least one operation to manage at least one deviation report; and based on determining that the at least one rule deviation is not to be implemented, performing at least one operation to manage the source code.