Homomorphic Encryption for Text Data Using Vector Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing encryption technologies require decryption for calculations, leading to resource wastage and potential data leakage, especially when dealing with unstructured data like text, images, and voices, as they only support homomorphic encryption for numerical data.
Innovation Solution
The method involves dividing text data into sentence units, calculating vector values using an encoding algorithm like BERT, and performing homomorphic encryption on these vector values, allowing for the generation of homomorphic encrypted messages that can be calculated without decryption, while storing text data securely by separating encrypted numerical and sequence indices into real and imaginary number areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional encryption technology is used, then data security is maintained during transmission, but decryption is required for calculations which wastes resources and time and creates security vulnerabilities
Solution Approach 1:
The patent changes the fundamental parameter of encryption from traditional symmetric/asymmetric encryption to homomorphic encryption, which allows calculations to be performed directly on encrypted data without decryption. This transformation enables the system to maintain data security while eliminating the time-consuming decryption step, as the encrypted data itself can be processed arithmetically to produce encrypted results that correspond to the correct plaintext operations.
2Reliability
If traditional encryption technology is used, then data security is maintained, but the other party must temporarily decrypt messages for calculation which creates security vulnerabilities
Solution Approach 1:
The patent transforms the encryption parameter to homomorphic encryption, which fundamentally changes the security model by allowing calculations on encrypted data without exposing the plaintext. This eliminates the security vulnerability where the other party must temporarily decrypt messages for calculation, as the homomorphic properties enable direct computation on ciphertext, producing encrypted results that never require intermediate decryption.
3Productivity
If homomorphic encryption is applied to numerical data only, then calculation without decryption is enabled, but unstructured data such as text, images, and voices cannot be processed
Solution Approach 1:
The patent changes the parameter of data representation by introducing vector encoding algorithms (such as BERT, Word2Vec, or GloVe) that transform unstructured data like text into numerical vectors. These vectors can then be subjected to homomorphic encryption and arithmetic operations. This parameter change enables the system to process unstructured data while maintaining the security and efficiency benefits of homomorphic encryption, as the vector representation bridges the gap between unstructured data and numerical computation.
Solution Approach 2:
The patent introduces vector encoding algorithms as an intermediary between unstructured data and homomorphic encryption. The encoding algorithm converts text, images, or voices into numerical vectors, which then serve as the input to homomorphic encryption. This intermediary enables the system to handle diverse unstructured data types while maintaining calculation efficiency and security, as the vectors provide a numerical interface that works with homomorphic operations.
4Productivity
If text data is converted to vector values, then homomorphic encryption can be applied, but the structure and meaning of original text may be lost
Solution Approach 1:
The patent changes the parameter of data representation to vector form using encoding algorithms that preserve semantic meaning. Advanced encoding algorithms like BERT maintain contextual and semantic information in the vector representation, allowing homomorphic encryption to be applied while retaining the essential meaning of the original text. The vector encoding transforms text into a numerical format that captures semantic relationships, enabling both encryption capability and information preservation.
Data Source
AI summary
A method of processing an encrypted message is provided. The method of processing an encrypted message includes dividing text data into sentence units, calculating a vector value of a predetermined size corresponding to each sentence by using a predetermined encoding algorithm for each sentence unit, and generating a homomorphic encrypted message by performing homomorphic encryption on the calculated vector value.