Sensitive Data Reclassification Using Metadata Confidence Rescoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional security management systems lack comprehensive computing logic and infrastructure to accurately evaluate data sensitivity, leading to deficiencies in identifying sensitive information, resulting in false positives and inadequate security posture management, which can lead to unauthorized access and malicious operations.
Innovation Solution
Implementing a sensitive data reclassification engine that evaluates scanned metadata to determine and update data classifications, using machine learning models to enhance data sensitivity confidence scores and improve the accuracy of identifying sensitive data items.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If heuristic approaches are used to determine data sensitivity, then the security management system can operate with simpler logic and infrastructure, but the accuracy of data classification deteriorates resulting in false positives and deficient functioning
Solution Approach 1:
The patent segments the data classification process into multiple stages: initial heuristic screening followed by metadata-based reclassification. This segmentation allows the system to use simple heuristic methods for broad filtering while applying more sophisticated metadata analysis only to borderline cases, thereby maintaining accuracy without requiring complex logic throughout the entire system.
Solution Approach 2:
The patent introduces metadata as an intermediary layer between the data item and the classification decision. Instead of directly classifying data items using complex logic, the system first extracts metadata attributes (file type, location, access patterns, etc.) which then serve as intermediate features for more accurate sensitivity evaluation, reducing the need for complex direct analysis.
2Ease of manufacture
If heuristic approaches are used for identifying sensitive data, then the system implementation is simpler, but the reliability of threat protection deteriorates leading to potential unauthorized access
Solution Approach 1:
The patent applies preliminary metadata extraction and analysis before final classification decisions are made. By pre-processing data items to gather relevant metadata attributes (file extensions, directory locations, access control lists, etc.), the system builds a foundation for more reliable classification without complicating the core implementation, as metadata collection is a standard operation in most systems.
Solution Approach 2:
The patent changes the parameters used for classification from simple heuristic rules to multi-dimensional metadata attributes. Instead of relying on single-factor heuristic decisions, the system evaluates multiple metadata parameters (file type, size, location, access patterns, modification history) simultaneously, significantly improving threat protection reliability while maintaining implementation feasibility through standardized metadata access.
3Measurement precision
If comprehensive metadata scanning and reclassification operations are implemented, then the accuracy of security posture information is improved, but the computing resources and processing time increase
Solution Approach 1:
The patent implements partial reclassification by applying metadata-based analysis selectively rather than to all data items uniformly. The system identifies data items with intermediate confidence scores from heuristic analysis and applies comprehensive metadata scanning only to these borderline cases, achieving high accuracy for critical decisions while avoiding unnecessary processing of clearly safe or clearly sensitive items, thus maintaining productivity.
Solution Approach 2:
The patent incorporates feedback mechanisms where classification results and metadata patterns are continuously analyzed to refine future classification decisions. The system learns from past classifications and metadata correlations, adjusting its metadata scanning depth and reclassification triggers based on observed patterns, thereby improving accuracy over time while optimizing resource usage by reducing redundant comprehensive scans.
Data Source
AI summary
Methods, systems, and computer storage media for providing data security posture management using a sensitive data reclassification engine in a security management system. Data security posture management provides security operations—including identifying and remediating risk exposure—to securely manage data, resources, and workloads. In operation, sensitive data scan results are accessed. A scanned data item having a first data sensitivity confidence score—and a first data classification that indicates that the scanned data item is a potentially sensitive data item—is identified in the sensitive data scan results. Based on the scanned data item having the first data classification, scanned metadata comprising a sensitive data attribute is accessed. Based on a second data sensitivity confidence score associated with the sensitive data attribute and the scanned data item, the first data classification is replaced with a second data classification that indicates that the scanned data item is a sensitive data item.