DOM-Based Sensitive Data Remediation System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tools for remediating sensitive information, such as personally identifiable information, are resource-intensive and lack flexibility, requiring significant processing time, memory resources, and power, with neural networks being trained for specific types of data and unable to adapt without retraining.
Innovation Solution
A system that uses rules applied to Document Object Model (DOM) structures to identify and remediate sensitive data, leveraging DOM structures for quick traversal and validation to achieve high accuracy with reduced processing time, memory resources, and power, while allowing for flexibility in identifying different types of sensitive data without retraining.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If neural networks are used to remediate sensitive information, then accuracy can be achieved, but processing time and resource consumption increase significantly
Solution Approach 1:
The patent segments the remediation process into distinct phases: DOM structure traversal to identify candidate elements, pattern matching to detect sensitive information, and validation to confirm sensitivity. This segmentation allows each phase to be optimized independently, replacing the monolithic neural network approach with efficient rule-based operations that reduce processing time while maintaining accuracy.
Solution Approach 2:
The patent substitutes the mechanical system of neural network computations with rule-based DOM traversal and pattern matching mechanisms. By using structured DOM queries and regex patterns instead of neural network inference, the system achieves the same accuracy goals with significantly reduced computational overhead and faster execution.
2Measurement precision
If neural networks are trained for specific types of data, then high accuracy is achieved, but adaptability to different data types is lost
Solution Approach 1:
The patent implements a universal rule-based framework that can handle multiple types of sensitive information (SSN, credit card numbers, bank account numbers, etc.) through a single system. The DOM traversal mechanism and pattern matching engine are designed to be data-type-agnostic, allowing the same infrastructure to adapt to different sensitive data types by simply changing the patterns and validation rules, eliminating the need for retraining.
Solution Approach 2:
The patent uses parameter changes to adapt to different data types. Instead of retraining neural networks, the system modifies runtime parameters such as regex patterns, validation rules, and DOM element selectors to match the specific type of sensitive data being processed. This allows rapid adaptation to new data types by changing configuration parameters rather than model weights.
3Reliability
If existing remediation tools are used, then sensitive information can be masked, but significant memory resources and power are consumed
Solution Approach 1:
The patent employs lightweight, disposable rule-based processing operations instead of heavy neural network models. Each DOM traversal and pattern matching operation is computationally inexpensive and can be discarded after use, consuming minimal memory and power. The system loads only the necessary patterns and validation rules for the current processing task, avoiding the memory footprint of pre-trained neural networks.
4Reliability
If existing remediation tools are used, then sensitive information can be remediated, but flexibility to adapt without retraining is limited
Solution Approach 1:
The patent implements a dynamic rule-based system where patterns, validation rules, and DOM selectors can be modified at runtime without retraining. The system dynamically adapts to different sensitive data types by loading appropriate configuration parameters and patterns, providing both reliability through validation and flexibility through easy reconfiguration.
Data Source
AI summary
In some implementations, a masking device may receive rules and a document object model (DOM) structure. Each rule may indicate a corresponding element, a corresponding pattern, and a type of remediation. The DOM structure may include elements, where each element is associated with text. The masking device may traverse the DOM structure to identify elements that map to corresponding elements indicated by the rules. The masking device may determine whether text, associated with the identified elements, is sensitive information by determining whether the text maps to corresponding patterns indicated by the rules. The masking device may perform validation on the sensitive information. The masking device may modify the DOM structure based on the sensitive information, the validation, and a type of remediation indicated by the rules. Accordingly, the masking device may output the modified DOM structure.