Homomorphic Encryption Text Search Without Query Decryption
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing encryption technologies waste resources and time on decryption, and expose personal information to risk of leakage during decryption for calculations, especially with the rise of deep learning models processing unstructured data like images and text.
Innovation Solution
A method using homomorphic encryption to process queries as ciphertext, converting text into vector values, generating query ciphertexts, and performing calculations without decryption, ensuring personal information remains protected.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional encryption/decryption technology is used to protect query security, then personal information can be protected during transmission, but resources and time are wasted during decryption operations and personal information may be leaked to third parties during temporary decryption for calculation
Solution Approach 1:
The patent introduces homomorphic encryption as an intermediary mechanism that enables calculations to be performed on encrypted data without decryption. The server operates on ciphertext directly, eliminating the need for decryption during processing, thus saving time and preventing information leakage while maintaining security.
Solution Approach 2:
The patent inverts the traditional encryption paradigm by enabling computations on encrypted data rather than requiring decryption first. Instead of encrypting → transmitting → decrypting → processing, the system transmits → processing (on ciphertext) → decrypting, thereby eliminating the security risk window during processing.
2Reliability
If traditional encryption/decryption technology is used to protect query security, then personal information can be protected during transmission, but computational resources are wasted during decryption operations and personal information may be leaked to third parties during temporary decryption for calculation
Solution Approach 1:
Homomorphic encryption serves as an intermediary that allows the server to perform computations directly on ciphertext without decryption. This eliminates the need for resource-intensive decryption operations while maintaining security, thereby reducing computational resource consumption.
Solution Approach 2:
The patent replaces the mechanical decryption-computation-encryption process with a homomorphic computation system that operates directly on ciphertext. This substitution eliminates the need for decryption operations, reducing computational overhead and energy consumption.
3Reliability
If homomorphic encryption is used to process queries as ciphertext without decryption, then personal information remains protected during calculation, but the complexity of the encryption system increases
Solution Approach 1:
The patent employs a universal homomorphic encryption framework that can handle various types of computations (similarity searches, text processing) on encrypted data without requiring different encryption schemes. This multi-functionality justifies the initial complexity by providing broad security benefits across multiple operations.
Solution Approach 2:
The system manages encryption complexity by carefully selecting and adjusting cryptographic parameters (security levels, ciphertext dimensions) to balance between security strength and computational feasibility. This parameter optimization reduces unnecessary complexity while maintaining robust protection.
4Reliability
If text is converted into vector values and processed through homomorphic encryption, then secure text search is enabled, but the processing time for encoding and encryption increases
Solution Approach 1:
The patent performs text encoding into vector values and homomorphic encryption in advance, before the search operation. This preliminary processing allows the actual search to be performed efficiently on pre-prepared ciphertext, reducing the perceived processing time for end users.
Solution Approach 2:
The system maintains continuous useful action by performing encoding and encryption operations in parallel where possible, and by keeping encrypted representations readily available for immediate search operations, thereby minimizing overall processing time despite the added security steps.
Data Source
AI summary
A text search method is disclosed. The text search method includes, based on a query including a text being input, computing a vector value having a preset size by using a preset encoding algorithm, the vector value corresponding to the text, generating a query ciphertext by homomorphic encryption for the computed vector value, transmitting the generated query ciphertext to a server, receiving a calculation result ciphertext having similarity information with the query for each of a plurality of indexes, determining an index having a preset similarity by restoring the calculation result ciphertext, and receiving information corresponding to the index by transmitting the determined index to the server.