Privacy Policy Recommendation Engine for Data Utility Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data sharing and analytics platforms face challenges in balancing privacy and data utility, as organizations struggle to share cybersecurity data while protecting sensitive information and maintaining the effectiveness of analytics, due to unclear anonymization requirements and rigid data policies.
Innovation Solution
A computer-implemented method using a Privacy Policy Recommendation Engine (PPRE) with a Restricted Boltzmann Machine (RBM) to automatically generate privacy policy recommendations, determining which data attributes need to be in non-anonymized formats for accurate analytics, thereby optimizing the balance between privacy and data utility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If data is anonymised to protect privacy, then privacy risk is reduced, but data utility for analytics deteriorates
Solution Approach 1:
The system dynamically determines the appropriate level of anonymization for each data attribute by analyzing analytics functions and their accuracy requirements. Instead of applying uniform anonymization, the system changes the anonymization parameter (level of anonymization) based on the specific analytics function's needs, allowing non-anonymized formats when high accuracy is required and anonymized formats when privacy protection is prioritized.
Solution Approach 2:
The patent applies different anonymization qualities to different data attributes based on their sensitivity and analytical value. Identity information receives complete anonymization or removal, while other attributes receive selective anonymization based on their role in specific analytics functions, creating local quality variations in the data treatment approach.
2Object-affected harmful factors
If manual privacy policy creation is used, then privacy protection can be customized, but process complexity and time consumption increase
Solution Approach 1:
The system enables self-service privacy policy generation by automatically analyzing analytics functions, data attributes, and their requirements. The privacy policy recommendation engine performs the complex analysis and policy determination autonomously, eliminating the need for manual policy creation while maintaining customized privacy protection tailored to each organization's specific analytics needs.
Solution Approach 2:
The system incorporates feedback mechanisms where analytics function requirements feed back into privacy policy determination. The analysis of analytics functions and their accuracy requirements provides continuous feedback that shapes the recommended privacy policies, ensuring they are both protective and analytically effective.
3Object-affected harmful factors
If data is modified to reduce privacy risk, then privacy protection improves, but analytics accuracy deteriorates
Solution Approach 1:
The system dynamically adjusts the anonymization parameter for each data attribute based on the specific analytics function's accuracy requirements. When high accuracy is needed, the system changes the parameter to non-anonymized format; when privacy protection is prioritized, it changes to anonymized format, optimizing the balance between these competing requirements.
Solution Approach 2:
The privacy policy is made dynamic rather than static, adapting to different analytics functions and their specific requirements. The system determines the appropriate data modification level in real-time based on the analytics function being executed, allowing the same data to have different privacy protection levels for different analytical purposes.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A computer-implemented method is provided for providing a privacy policy recommendation, the method comprising steps of: retrieving one or more analysis results, each of the one or more analysis results being generated by analysing an input dataset comprising one or more data attributes by performing one or more analytics functions, and each of the one or more data attributes having an anonymised format or a non-anonymised format; determining, for each of the analytics functions, the data attributes of a first type, the data attributes of the first type being a set of the one or more data attributes that need to have non-anonymised formats in order for the corresponding analysis result to have an acceptable level of accuracy and/or quality; and generating the privacy policy recommendation indicating, for each of the analytics function, that the data attributes of the first type need to be in non-anonymised formats. A corresponding computer system and computer program are also provided.