De-Identified Content Classification Through Hypernym Relationships
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies struggle to effectively classify personal data from de-identified content, particularly in compliance with regulatory requirements such as GDPR, due to the complexity of identifying and annotating sensitive information within structured documents.
Innovation Solution
Applying predefined rules based on document structure to associate text with hypernyms, using hypernym discovery techniques, and applying causality rules to identify and classify personal data, with the aid of deep neural networks and structural context vectors, to annotate de-identified content as belonging to a class of protected content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If de-identification techniques are applied to protect personal data, then privacy protection is improved, but the ability to accurately classify and identify personal data for compliance purposes deteriorates
Solution Approach 1:
The patent introduces structural elements and hypernym relationships as intermediary layers between the original text and the de-identified content. These intermediaries preserve the semantic structure and contextual relationships that enable accurate classification while allowing the actual personal data to be de-identified. The structural elements act as a bridge that maintains classificability without requiring the original personal data to be visible.
Solution Approach 2:
The system performs preliminary classification and annotation of structural elements before de-identification is applied. By establishing hypernym relationships and identifying potential personal data categories in advance, the system creates a framework that guides subsequent classification efforts. This preliminary structuring enables accurate identification even after de-identification techniques modify the original content.
2Measurement precision
If complex classification rules are applied to identify personal data, then classification accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent segments the classification task into multiple independent components: structural element identification, hypernym relationship determination, and causal rule application. Each component can be processed separately and in parallel, reducing overall processing time. The segmentation allows the system to apply different levels of complexity to different parts of the document based on their classification needs.
Solution Approach 2:
The system applies different classification rules and levels of analysis to different portions of the document based on local characteristics. Structural elements with higher importance or those containing more sensitive information receive more rigorous analysis, while less critical portions use simpler classification approaches. This localized quality adjustment optimizes the balance between accuracy and processing efficiency.
Data Source
AI summary
Methods, systems, and computer program products for classifying data from de-identified content are provided herein. A computer-implemented method includes applying one or more rules to identify one or more structural elements of a document; determining, based at least in part on the one or more structural elements, one or more pairs of words within the document having a hypernym relationship; extracting de-identified content within the document based on one or more de-identification techniques applied to the document; and applying a set of causal rules to the de-identified content and the one or more pairs of words to annotate at a least a portion of the de-identified content as belonging to a class of protected content.