NLP to Database Bridge Adapter for Query Translation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

There is a need for improved technologies to provide flexible and easy-to-configure bridges between natural language processing (NLP) tools and database tools, as existing approaches require significant engineering effort and are often inadequate for multiple database deployments.

Innovation Solution

The disclosed technologies provide a bridge between the output of an NLP engine, in the form of a structured representation of a natural language query, and the exposed interface of a database query processor. This bridge extracts a sub-intent, maps it to a specific database interface, and transforms query parameters to conform with the target database interface, allowing for the construction of a database query.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If custom NLP tools are built for specific database deployments, then the NLP tool can be tailored to meet specific requirements, but the engineering effort and time required increase significantly

Engineering Contradiction:
ImproveNLP tool customizationVSAvoidEngineering effort
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent creates a universal adapter layer that enables a single NLP tool to work with multiple database systems through standardized interfaces. The adapter translates between different NLP output formats and database-specific query formats, eliminating the need to build custom NLP tools for each database deployment while maintaining full adaptability to specific requirements

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent introduces an adapter as an intermediary component between the NLP tool and the database system. This adapter handles the translation and compatibility issues, allowing the NLP tool to remain generic while still meeting specific database deployment requirements through configurable adapter implementations

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If database tools are customized to match NLP outputs, then compatibility is improved, but the effort must be repeated when changing NLP tools

Engineering Contradiction:
ImproveDatabase tool compatibilityVSAvoidRe-engineering effort
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

Instead of customizing database tools to match NLP outputs, the patent inverts the approach by creating adapters that translate NLP outputs into database-compatible formats. This reversal places the adaptation logic in the NLP side rather than the database side, making it easy to switch NLP tools without re-engineering database tools

Inventive Principle:
Principle #13The other way round (Inversion)

Solution Approach 2:

The adapter framework provides universal compatibility with multiple NLP tools through a standardized interface. Each adapter is designed to work with specific NLP tools, but the overall system maintains the ability to switch between different NLP tools without modifying the database tools themselves

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Productivity

If NLP tools are built for specific database deployments, then performance can be optimized for that deployment, but the solution becomes inadequate for other database deployments

Engineering Contradiction:
ImproveQuery processing performanceVSAvoidDatabase deployment flexibility
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent creates a dynamic adapter system that can be configured and adjusted for different database deployments. The adapter layer allows the system to optimize performance for specific deployments through configuration rather than hard-coded customization, maintaining both high performance and broad adaptability across different database systems

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3945431B1Bridge from natural language processing engine to database engine
Publication Date: 2025.05.07 SAP SE
  • EP3945431B1 patent drawingFigure 1
  • EP3945431B1 patent drawingFigure 2
  • EP3945431B1 patent drawingFigure 3

AI summary

Methods and systems are disclosed for bridging a natural language processing (NLP) engine to a database engine, enabling natural language queries to be handled by the database engine. The NLP engine produces a query structure, containing an intent and one or more query parameters, for a natural language query. Rules are successively applied to the query structure. When a rule matches, a corresponding action is performed to identify a sub-intent, transform query parameters, or generate a filter string. The sub-intent is mapped to a query destination, which can be an intrinsic or extrinsic service offered by the data engine. A query string is constructed, including the query destination and any filter strings, and transmitted to the database engine to answer the original query. The disclosed technologies are flexible, easily reconfigurable, and future-proof. Variations and use cases are disclosed.