Security Classification by Machine Learning Models

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current systems face challenges in accurately and efficiently determining security classifications for electronic documents and setting appropriate access controls, particularly in large networks with sensitive information, due to the complexity of analyzing vast amounts of data and requiring manual intervention.

Innovation Solution

The system employs machine learning logic to generate classification models based on content, metadata, and contextual factors, using weight parameters to determine security classifications and adjust access privileges, and automatically retrain models to maintain accuracy in the face of data changes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual methods are used to determine security classifications for electronic documents, then accuracy can be maintained through human judgment, but productivity is reduced due to the time-consuming nature of manual analysis

Engineering Contradiction:
Improveclassification accuracyVSAvoidclassification speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The system enables electronic documents to classify themselves automatically by analyzing their own content, metadata, and contextual factors through machine learning models, eliminating the need for manual human review while maintaining classification accuracy

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

Manual human judgment for security classification is replaced with automated machine learning systems that process document content, metadata, and contextual information to determine appropriate security classifications, dramatically increasing productivity

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If comprehensive analysis of content, metadata, and contextual factors is performed to improve classification accuracy, then measurement precision is improved, but device complexity increases due to multiple inference models and weight parameters

Engineering Contradiction:
Improveclassification accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The classification system is divided into separate inference models for different aspects: content analysis, metadata analysis, and contextual factor analysis. Each model specializes in one aspect and can be independently trained and optimized, reducing overall system complexity while maintaining comprehensive analysis

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The machine learning framework provides a universal platform that handles multiple classification aspects (content, metadata, context) through a unified architecture with weight parameters that can be adjusted based on the importance of different factors, simplifying the integration of multiple analysis functions

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Measurement precision

If machine learning models are retrained frequently to maintain accuracy with changing data, then measurement precision is maintained, but loss of time occurs during the retraining process

Engineering Contradiction:
Improvemodel accuracyVSAvoidretraining time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system implements continuous feedback loops where classification performance is monitored and automatically triggers retraining when accuracy thresholds are not met or when new contextual factors are identified, ensuring models remain accurate without excessive retraining

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

Instead of continuous retraining, the system employs periodic retraining cycles triggered by specific conditions such as accumulation of new data, changes in security policies, or performance degradation, reducing time loss while maintaining model accuracy

Inventive Principle:
Principle #19Periodic action

Data Source

PatentEP3355248B1Security classification by machine learning
Publication Date: 2024.12.18 ACCENTURE GLOBAL SOLUTIONS LTD
  • EP3355248B1 patent drawingFigure 1
  • EP3355248B1 patent drawingFigure 2
  • EP3355248B1 patent drawingFigure 3

AI summary

A computer-implemented method that includes a computing system receiving a plurality of data items that include content data. The system then generates a first inference model based on the received content data. The system also receives metadata associated with respective data items and then generates a second inference model based on the received metadata data. The system further determines, based on the first inference model, a first weight parameter associated with the content data, and determines, based on the second inference model, a second weight parameter associated with the metadata. The system then generates a classification model for determining a security classification of at least one data item of the plurality of data items. The classification model is generated based on at least one of the first weight parameter or the second weight parameter.