Zero-Knowledge Document Structure and Content Separation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud-based document processing systems expose sensitive content to central servers, compromising user privacy and system security as they are not 'zero-knowledge' systems.
Innovation Solution
A method that separates a document's structure from its content, encrypting the content and generating tokens to replace it, allowing only the server to convert formats while keeping encryption keys local, ensuring the server has 'zero-knowledge' of the document's data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If cloud-based document processing is implemented with central server conversion, then document format conversion and real-time collaboration are enabled, but user privacy and system security deteriorate due to content exposure to the server
Solution Approach 1:
The document is segmented into two distinct parts: structure (metadata) and content (data). The structure portion containing formatting, layout, and organizational information is sent to the cloud server in plaintext for conversion operations. The content portion containing sensitive information is encrypted locally and never exposed to the server. This segmentation allows the server to perform format conversion on the structure while the encrypted content remains protected, resolving the contradiction between enabling cloud-based conversion and maintaining security.
Solution Approach 2:
The sensitive content is extracted from the document and replaced with tokens or placeholders before being sent to the cloud server. The server only receives and processes the structural metadata without accessing the actual content. After conversion, the encrypted content is recombined with the converted structure on the client side. This extraction ensures the server has zero knowledge of the content while still enabling format conversion functionality.
2Reliability
If document content is encrypted locally before upload, then user privacy and system security are improved, but server-based format conversion capability may deteriorate due to inability to process encrypted data
Solution Approach 1:
The document processing system segments the document into structure and content components. The structure portion is sent to the server in plaintext, allowing the server to perform format conversion operations. The content portion is encrypted locally and sent separately or embedded as tokens within the structure. This segmentation enables the server to convert formats while the encryption protects content, resolving the contradiction between security and conversion capability.
Solution Approach 2:
Tokens or placeholders serve as intermediaries between the encrypted content and the cloud server. The server operates on these tokens within the document structure without needing to decrypt or understand the actual content. The tokens act as mediators that allow format conversion to occur on structural elements while the real content remains encrypted and protected, enabling both security and conversion functionality.
Data Source
AI summary
Techniques disclosed herein decouple a document's structure from its general content wherein the structure is retained in plaintext (both at a client device and in a server system) and the data is retained in cyphertext, and where the cloud-based server system is not tasked with the saving or management of the relevant cryptographic keys. Because the network- or cloud-based server system has “zero-knowledge” about the document's data content or the relevant cryptographic keys, an attack on the server system does not put the security of the document's data at risk. In addition, the network- or cloud-based server system may be used to perform the computationally intensive tasks of converting the document between a first format (often associated with a full-function document processing application not supported by the client device) and a second format (easily displayed and manipulated by a client device).


