ML-Based Unstructured Data Anonymization for PII Disclosure Risk
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to effectively anonymize unstructured data due to the inability to detect and assess the risk of personally identifiable information (PII), leading to potential disclosure uncertainties and compliance issues.
Innovation Solution
A system utilizing machine-learning models to detect PII in unstructured data, simulate undetected PII, and apply transformations or resyntheses to mitigate disclosure risks, ensuring compliance through auditable proof of efficacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine-learning models are used to detect PII in unstructured data, then detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The system performs preliminary actions by simulating undetected PII before final risk assessment. This allows the model to account for potential missed detections in advance, improving overall detection accuracy without requiring exponentially more computational resources during the main processing phase.
Solution Approach 2:
The system introduces an intermediary simulation component that generates hypothetical undetected PII instances. This intermediary layer bridges the gap between detected and undetected PII, allowing the risk assessment model to evaluate disclosure risk more comprehensively without directly analyzing every possible undetected attribute.
2Reliability
If simulation of undetected PII is performed, then risk assessment completeness is improved, but processing time increases
Solution Approach 1:
The system applies partial action by simulating only a representative subset of undetected PII rather than exhaustively generating all possible undetected instances. This simulation approach provides sufficient risk assessment completeness while avoiding excessive processing time that would result from complete enumeration of all potential undetected attributes.
3Object-affected harmful factors
If transformations are applied to mitigate disclosure risk, then data privacy protection is improved, but data utility is reduced
Solution Approach 1:
The system applies local quality by performing targeted transformations only on specific detected PII attributes that pose disclosure risks, rather than uniformly transforming all data. This selective approach protects privacy where needed while preserving data utility in non-sensitive areas.
Solution Approach 2:
The system utilizes parameter changes by applying different transformation techniques (such as generalization, suppression, or perturbation) to different PII attributes based on their specific risk characteristics. This allows optimization of the balance between privacy protection and data utility for each attribute type.
4Measurement precision
If comprehensive risk assessment is performed on all attributes, then disclosure risk measurement is improved, but computational resources are excessive
Solution Approach 1:
The system extracts and focuses computational resources on assessing risk for detected PII attributes and representative simulated undetected attributes, rather than uniformly processing all attributes. This extraction of critical assessment targets maintains measurement precision while reducing overall computational resource consumption.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for anonymizing unstructured data. In some implementations, a server can receive unstructured data. The server can automatically detect attributes in the unstructured data using a trained machine-learning model and can determine an amount of undetected attributes and detected attributes in the unstructured data. The server can simulate additional attributes for the unstructured data according to the amount of undetected attributes. The server can analyze a risk of disclosure in the unstructured data using the detected attributes and the simulated additional attributes. The server can modify the detected attributes according to the analyzed risk of disclosure and replace the detected attributes with the modified detected attributes in the unstructured data.