Role-Based LLM Document Retrieval Without Sensitive Data Exposure
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional search tools fail to distinguish between user roles within an organization, leading to the exposure of sensitive information to users who are not authorized to access it.
Innovation Solution
Employing artificial intelligence (AI) large language models (LLMs) to selectively extract and display information from a knowledge base based on a user's role, using a user interface that allows for conversational input and semantic search to identify relevant blocks of text, while filtering out sensitive content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional search tools retrieve all relevant information from the knowledge base, then the completeness of information retrieval is improved, but the risk of exposing sensitive information to unauthorized users increases
Solution Approach 1:
The system performs preliminary classification of information into public and sensitive categories before retrieval. Role-based access control rules are pre-established and applied during the search process to filter sensitive information before it can be exposed to unauthorized users, thus maintaining information completeness for authorized users while preventing leaks to unauthorized users
Solution Approach 2:
An intermediary access control module is introduced between the search tool and the knowledge base. This intermediary automatically identifies user roles, applies appropriate access rules, and filters information based on authorization levels. It acts as a mediator that allows complete information retrieval for authorized users while blocking sensitive information for unauthorized users without requiring users to manually filter content
2Ease of operation
If the search tool displays full documents to satisfy user requests, then the satisfaction of user information needs is improved, but the precision of information delivery decreases due to inclusion of unauthorized content
Solution Approach 1:
The system applies different quality filters to different parts of the retrieved information based on user authorization. Instead of uniformly filtering all information or none, it selectively applies access control rules to specific documents or sections within documents, allowing authorized content to be displayed in full while filtering only the unauthorized portions, thus maintaining high information satisfaction with precise delivery
3Reliability
If the system implements role-based access control filtering, then the security of information is improved, but the complexity of the search system increases
Solution Approach 1:
The access control system automatically retrieves user role information from existing user profiles and databases without requiring manual intervention. The system self-configures access rules based on pre-defined role templates and automatically applies appropriate filters during search operations, reducing the need for complex manual configuration and ongoing system management while maintaining high security standards
Data Source
AI summary
Certain aspects of the disclosure provide a method that selectively extracts information from a knowledge base of an organization based on roles of users in the organization. A first large language model (LLM) is used to identify one or more blocks of text that a user is permitted to access based on a prompt input by the user via a user interface (UI) and the role of the user in the organization. A second LLM is used to generate a response to the prompt based on text of the prompt and the one or more blocks of text that the user is permitted to access. The response is displayed is UI. The response contains the one or more blocks of text that the user is permitted to access in the knowledge base.