Vector Query Operators for Relational Database Index and Search

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Relational databases lack the capability to efficiently index, maintain, and search vector embeddings due to their operators not being designed for such data, leading to performance issues and the need for time-consuming custom programming.

Innovation Solution

Convert vector-related requests into logical operator trees using query operators within a relational database system, enabling vector operations without vector-specific executable code, leveraging existing database query operators to support vector search and indexing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If relational databases use traditional query operators designed for tabular data, then the database structure remains simple and well-established, but the capability to efficiently index and search vector embeddings is lost

Engineering Contradiction:

Improvevector search capabilityVSAvoidquery operator complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent extends the functionality of existing relational database query operators to handle vector embeddings. The same query operators that work with tabular data are made universal to also process vector operations, allowing a single database system to handle both traditional relational queries and vector search operations without requiring separate specialized systems.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent changes the data type parameters of existing query operators to accommodate vector embeddings. By modifying how operators handle data types (from purely tabular to supporting both tabular and vector types), the system gains vector search capability while reusing the same operator framework, thus avoiding complete system redesign.

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If custom programming is implemented to add vector search functionality to relational databases, then vector search capability is achieved, but the implementation time and development effort increase significantly

Engineering Contradiction:

Improvevector search capabilityVSAvoidimplementation time

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent enables relational databases to perform vector search operations using their own existing query operator infrastructure. Instead of requiring external custom programming, the database system serves its own vector search needs by extending its native operators to handle vector data types, thus eliminating the need for time-consuming custom implementation.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent merges vector search functionality with the existing relational database query operator framework. By combining vector operations with traditional SQL operators into a unified system, the patent eliminates the need for separate custom programming implementations, reducing development time while achieving versatile search capability.

Inventive Principle:

Principle #5Merging (Combining)

3Adaptability or versatility

If vector-specific executable code and operators are used, then vector search functionality is achieved, but the integration with existing database components becomes complex and seamless operation is lost

Engineering Contradiction:

Improvevector operation supportVSAvoidsystem integration complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent makes existing database query operators universal by enabling them to handle both traditional tabular data and vector embeddings. This eliminates the need for separate vector-specific operators and their associated integration complexity, as the same operators seamlessly handle different data types based on context.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

Instead of creating vector-specific operators that would require integration with the existing database system, the patent inverts the approach by extending the existing operators to handle vector data. This reversal simplifies integration because the extension builds upon the already-integrated query operator framework rather than requiring new components to be integrated.

Inventive Principle:

Principle #13The other way round (Inversion)

Data Source

PatentUS12536186B1Implementing vector index build and search using query operators

Publication Date: 2026.01.27 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

Methods, systems, and computer program products are provided that implement vector-related requests using query operators. For example, a system includes a database, a parser, a converter, an optimizer, and an execution engine. The database is associated with query operators that are non-vector-specific. The database stores a table with vector embeddings. The parser is configured to parse a request indicating a vector operation associated with the vector embeddings. The converter is configured to convert the vector operation into a logical operator tree comprising a representation of the request as a logical flow of the query operators, enabling the vector operation without vector-specific executable code or operators. The optimizer is configured to convert the logical operator tree into an executable plan. The execution engine is configured to execute the executable plan against the table with vector embeddings.