Client-Side Query Preprocessing for AI Privacy Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hyperscale AI models pose a risk of information exposure as they transmit personal and sensitive information to servers during query processing.
Innovation Solution
A method and apparatus for providing an AI-based service that preprocesses user queries by replacing personal information with replacement words, transmits the preprocessed queries to a service server, and restores the original information in the response received from the server.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If user queries containing personal information are transmitted directly to the AI service server, then the AI service can be provided accurately, but personal information is exposed and leaked
Solution Approach 1:
The system performs preliminary action by preprocessing the user query on the client side before transmission. Specifically, it identifies and replaces personal information words in the query with placeholder tokens, so that when the query reaches the server, the sensitive information has already been protected. This preliminary masking action prevents information exposure while preserving the query's semantic meaning for accurate AI processing.
Solution Approach 2:
The system introduces an intermediary mechanism in the form of placeholder tokens that mediate between the user's personal information and the AI service. Instead of transmitting actual personal information, the client replaces sensitive words with intermediate tokens that carry positional and semantic information without exposing the original data. The server processes these tokens and returns results that are then mapped back to the original information context.
2Object-affected harmful factors
If personal information is replaced with replacement words before transmission, then information exposure is prevented, but the complexity of the system increases due to preprocessing and postprocessing requirements
Solution Approach 1:
The system extracts personal information from the user query before transmission by identifying sensitive words and separating them from the rest of the query. The extracted personal information is replaced with placeholder tokens, while the structural and semantic information is preserved. This extraction approach isolates the sensitive data handling to a specific module, making the overall system more manageable despite the added complexity.
Solution Approach 2:
The system creates a copy of the query structure with placeholders instead of original personal information. The placeholder tokens serve as copies that maintain the positional and contextual information needed for accurate processing, while the actual sensitive data remains on the client side. This copying mechanism allows the server to process the query structure without accessing the original sensitive information.
3Object-affected harmful factors
If preprocessing is performed on the client apparatus to replace personal information, then privacy is protected, but the processing time and computational resources increase
Solution Approach 1:
The system applies partial action by selectively replacing only the personal information words in the query rather than processing or masking the entire query. The placeholder replacement focuses specifically on sensitive tokens while leaving the rest of the query structure intact. This partial processing approach minimizes the time and computational resources required compared to complete query rewriting or encryption.
Data Source
AI summary
Provided is a method of providing an artificial intelligence (AI)-based service without exposure of personal information, which includes receiving, by a client apparatus, a user query, performing, by the client apparatus, preprocessing on the user query to replace a target word corresponding to personal information in the user query with a corresponding replacement word, transmitting, by the client apparatus, the preprocessed user query to a service server for providing an AI-based service, and replacing, by the client apparatus, the replacement word in an answer corresponding to the user query that is received from the service server with the target word corresponding thereto.