Cognitive Insight Security Classification via ML Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cognitive systems face security challenges as generated insights may require higher security classification than their individual data inputs, leading to potential unauthorized access and security flaws in data repositories.
Innovation Solution
A method using a trained machine learning model to analyze cognitive insights and adjust the security levels of both the insights and underlying data sources, ensuring that only authorized users can access the insights by assigning a higher security level if necessary, thereby maintaining data integrity and security.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the security level of cognitive insights is determined solely by the highest security level of individual data inputs, then the system is simple to operate, but the security classification may be insufficient and lead to unauthorized access
Solution Approach 1:
The system performs preliminary analysis of the cognitive insight content before final security classification, examining the actual information generated rather than just relying on input data classifications. This preliminary action ensures accurate security determination while maintaining operational simplicity through automated analysis.
Solution Approach 2:
The patent replaces manual security classification mechanisms with an automated machine learning model that analyzes cognitive insights and determines appropriate security levels. This substitution eliminates complex manual security level determination while improving classification accuracy through intelligent automated assessment.
2Reliability
If the security level of data sources is increased to match the highest possible insight security level, then security is improved, but access to legitimate users is restricted and system productivity decreases
Solution Approach 1:
The system applies different security levels to different data sources based on their specific contribution to each cognitive insight, rather than uniformly applying the highest security level to all data sources. This localized approach maintains data security while allowing legitimate users to access and generate insights from appropriately classified data sources.
Solution Approach 2:
The security level assignment is dynamic rather than static, adjusting security classifications based on the specific cognitive insight being generated and the actual security requirements of each data source. This dynamic approach prevents unnecessary security restrictions that would reduce productivity while maintaining appropriate security levels.
3Reliability
If manual review and adjustment of security levels is performed, then security accuracy is improved, but the time required for insight generation and access increases
Solution Approach 1:
The system performs self-service security level determination through automated machine learning models that analyze cognitive insights and assign appropriate security levels without requiring manual review. This self-service mechanism maintains high security accuracy while eliminating time delays associated with manual security level adjustments.
Solution Approach 2:
The system implements automated feedback loops where the machine learning model continuously analyzes generated insights, determines security levels, and adjusts data source classifications accordingly. This automated feedback process ensures accurate security classification without manual intervention, maintaining both security accuracy and operational efficiency.
Data Source
AI summary
Techniques for ensuring the security of cognitive insights are disclosed. A request to generate a cognitive insight is received from a requestor. The requestor is associated with a requestor data security level. The cognitive insight is generated using a first machine learning model and a plurality of data sources, each data source associated with a respective data security level. An insight data security level for the generated cognitive insight is identified based on the insight and the plurality of data sources. A first data security level associated with a data source of the plurality of data sources is modified, based on the identified insight data security level. It is determined, based on the requestor data security level and the insight data security level, that the requestor is authorized to access the generated insight. In response the generated insight is provided to the requestor.