Application Database Access Via Parameterized Secure Views
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current natural language to structured query language (NL2SQL) techniques fail to detect unauthorized access attempts in application-specific data, compromising data security by allowing users to manipulate large language models (LLMs) into providing unauthorized access.
Innovation Solution
Implementing parameterized secure view elements that use unique identifiers to generate virtual databases containing only accessible data for each user, coupled with LLMs for secure NL2SQL translation, ensuring users can only access permitted data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If NL2SQL translation is used to enable natural language access to databases, then user accessibility and ease of operation are improved, but data security and protection against unauthorized access deteriorate
Solution Approach 1:
The patent introduces a parameterized secure view element as an intermediary layer between the NL2SQL translation system and the actual database. This secure view acts as a mediator that receives SQL queries from the translation layer and filters them through user authorization checks before accessing the database, thereby enabling natural language access while preventing unauthorized data retrieval
Solution Approach 2:
The system performs preliminary authorization checks by evaluating user tokens against the required access permissions before executing any data retrieval operation. The parameterized secure view element pre-determines whether a user is authorized to access specific data before the actual query execution, blocking unauthorized access attempts at the translation stage
2Object-affected harmful factors
If parameterized secure view elements are implemented to enforce access control, then data security is improved, but system complexity increases
Solution Approach 1:
The patent creates a virtual copy of the database schema through parameterized secure views, which presents a simplified interface to authorized users. Instead of modifying the actual database structure or implementing complex access control mechanisms at the database level, the system creates a virtual layer that copies and filters data based on user tokens, reducing implementation complexity while maintaining security
Data Source
AI summary
The technology generally relates to securing end user access to application databases. An application receives a natural language query from an application end user requesting particular data from a database. The application identifies one or more tokens associated with the application end user to identify the application-specific data that is accessible to the application end user. The application provides the one or more tokens to one or more parameterized secure view elements. The parameterized secure view elements use the one or more tokens to create a virtual database that contains only the application-specific data that is accessible to the application end user. The application uses one or more machine learning models to translate the natural language query into a database query that the application uses to access the virtual database and retrieve the particular data to respond to the natural language query.