Sensitive Term Detection in Code via Vectorization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge lies in efficiently identifying and removing sensitive terms from computer code, which is a time-consuming and difficult process, especially when they are embedded within large lines of code, making quality control reviews cumbersome and delaying product completion.
Innovation Solution
A system and method that utilize machine learning techniques to predict the presence of sensitive terms in text data by converting the data into vector representations based on position, assignment, keyword, and threshold bearers, allowing for the identification and potential modification of sensitive terms within the code.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If quality control review is performed manually to identify sensitive terms in code, then detection accuracy can be maintained, but the process becomes time-consuming and delays product completion
Solution Approach 1:
The system performs preliminary detection of sensitive terms during code development using automated machine learning models. By detecting sensitive terms early in the development process rather than during final quality control review, the system maintains detection accuracy while significantly reducing the time required for product completion. The model analyzes code in real-time or near-real-time, allowing developers to correct issues before they reach the quality control stage.
Solution Approach 2:
The patent replaces manual quality control review (mechanical human process) with an automated machine learning system. The ML model processes code to identify sensitive terms, substituting the time-consuming manual review process while maintaining or improving detection accuracy. This automation eliminates the trade-off between manual detection accuracy and time consumption.
2Productivity
If sensitive terms are embedded within large lines of code, then code functionality can be maintained, but identification and removal becomes difficult and time-consuming
Solution Approach 1:
The machine learning model acts as an intermediary between the developer and the code, automatically analyzing large lines of code to identify embedded sensitive terms. The model processes the entire code context and highlights specific sensitive terms, making them easy to locate and remove. This intermediary system eliminates the difficulty of manually searching through large codebases while preserving code functionality.
Solution Approach 2:
The system creates a representation or analysis of the code structure to identify sensitive terms without modifying the original code during the detection phase. The ML model processes code copies or analyses, allowing thorough examination of large codebases without affecting the original development workflow. This copying approach enables efficient detection while maintaining code integrity.
3Reliability
If comprehensive quality control process is implemented to identify sensitive terms, then security can be improved, but the process becomes cumbersome and complex
Solution Approach 1:
The machine learning system performs self-service by automatically detecting and flagging sensitive terms without requiring complex manual quality control processes. The model independently analyzes code, identifies sensitive terms, and provides recommendations for remediation. This self-service capability maintains high security standards while eliminating the need for cumbersome multi-step quality control procedures.
Solution Approach 2:
The ML-based quality control system serves multiple functions: detecting sensitive terms, analyzing code context, providing remediation suggestions, and integrating with development workflows. This universal system replaces multiple separate quality control tools and processes with a single unified solution, maintaining comprehensive security coverage while reducing overall process complexity.
Data Source
AI summary
Methods, systems, and apparatuses are provided for predicting if an item of text data includes one or more sensitive terms. An item of text data may be received by a computing device from a second computing device associated with a user. The computing device may determine one or more words within the item of text data. An arrangement of the one or more words may be determined. The arrangement of the one or more words may be based on an assignment bearing vectorization of the item of text data and/or a position bearing vectorization of the item of text data. Based on the arrangement of the one or more words within the item of text data, the computing device may determine a potential sensitive term within the item of text data. The computing device may cause an output of an indication of the potential sensitive term.