Confidential Information Search Using Vectorized Metadata Access
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing database search methods are impractical for large volumes of confidential information objects due to the need-to-know principle, leading to inefficient and cumbersome search processes that expose users to irrelevant data and violate confidentiality.
Innovation Solution
Vectorizing content data of information objects through text, image, audio, and video embedding, and storing them as vectors in a database, allowing for context-sensitive searches and adherence to the need-to-know principle by ensuring only authorized users access relevant data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the need-to-know principle is applied to restrict access to confidential information objects, then security and confidentiality are improved, but search process complexity and user difficulty increase
Solution Approach 1:
The patent introduces metadata as an intermediary layer between the user and confidential content data. Users search and interact with metadata (which may be accessible), and the system mediates access to content data based on need-to-know principles. This intermediary approach maintains confidentiality while simplifying the search process, as users don't need to navigate complex access restrictions for every search operation.
Solution Approach 2:
The patent segments information objects into metadata and content data components with different access levels. Metadata can be searched and displayed according to user permissions, while content data access is separately controlled based on need-to-know principles. This segmentation allows the search interface to remain simple and accessible while maintaining strict security controls on the actual confidential information.
2Productivity
If users are exposed to large volumes of confidential information objects during search, then search completeness is improved, but security risks and exposure to irrelevant data increase
Solution Approach 1:
The patent applies local quality by differentiating access permissions for different parts of the search system. Metadata can be broadly accessible to ensure search completeness, while content data access is restricted to specific users based on their need-to-know permissions. This allows the system to maintain both search completeness and security by applying different quality levels of access control to different components.
Solution Approach 2:
The system uses metadata as an intermediary that allows users to conduct comprehensive searches without directly exposing confidential content data. The metadata layer enables search completeness while the mediation mechanism filters and controls access to actual content based on security requirements, reducing exposure to irrelevant and confidential data simultaneously.
3Ease of operation
If conventional keyword search methods are used on confidential documents, then search simplicity is maintained, but search accuracy and relevance assessment deteriorate
Solution Approach 1:
The patent changes the search parameter from simple keyword matching to vector-based semantic similarity comparison. Instead of relying on exact keyword matches, the system converts both queries and metadata into vectors and measures similarity through vector operations. This parameter change maintains ease of operation (users still perform simple searches) while dramatically improving measurement precision (relevance assessment through semantic understanding).
Data Source
Figure 1
Figure 2
AI summary
In a method for searching confidential information objects (5), an electronic platform (1) comprises a computing unit (2) and a database (3). Use of the platform (1) requires user access authorization (4), and the platform (1) provides access to a multitude of information objects (5). A search query (7) from a searching user (4, 21) is transmitted to the computing unit (2) and triggers a search in the database (3). The information objects (5) comprise content data (17) and metadata (8). At least some, and preferably all, of the content data (17) of an information object (5) are vectorized by embedding text, audio, and/or video and stored as a vector (9) in the database (3).